DSSM
Uploaded by
Bharath RamDSSM
Uploaded by
Bharath RamDeterministic & Stochastic Statistical Methods
Unit – I
Data Representation
Covariance: Covariance is a measure of the relationship between two random variables.
Let X and Y be two random variables with means X and Y respectively. Then the covariance
between two variables is defined by the relation Cov X,Y E X X Y Y or
Cov X,Y E XY E X E Y
Proof: We have Cov X,Y E X X Y Y
Cov X,Y E X EX Y EY
Cov X,Y E XY XE Y YE X E X E Y
E XY E Y E X E X E Y E X E Y
Cov X,Y E XY E X E Y
Note:
• Covariance of two independent variables is zero. But converse need not be true.
• If a and b are constants
Cov(X + a,Y + b) = Cov(X,Y)
Cov(aX,bY) = abCov(X,Y)
1 n
• Population Covariance of X and Y is given Cov X,Y xi x y i y
Ni 1
1 n
• Sample Covariance of X and Y is given Cov X,Y xi x y i y
N 1i 1
• Cov X,X Var X and Cov Y,Y Var Y
DSSM Compiled by: Magisetty Harikrushna 1
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Find the covariance, covariance matrix and correlation of the following data
01. 02. 03.
X 2.1 2.5 4.0 3.6 X 98 87 90 85 95 75 X 2 5 6 8 9
Y 8 12 14 10 Y 15 12 10 10 16 7 Y 4 3 7 5 6
04. 05.
X 65.21 64.75 65.56 66.45 65.34 X 5 6 8 11 4 6
Y 67.15 66.29 66.2 64.7 66.54 Y 1 4 3 7 9 2
06.
X 13 15 17 18 19
Y 10 11 12 14 16
2 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Solution: Given data
X 2.1 2.5 4.0 3.6
Y 8 12 14 10
Here number of features n = 2
Number of samples N = 4
1 1
Computation table: Mean values x = ( x i ) and y = ( yi )
N N
x y ( x i − x ) ( x i − x )2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
2.1 8 -0.95 0.903 -3 9 2.85
2.5 12 -0.55 0.303 1 1 -0.55
4 14 0.95 0.903 3 9 2.85
3.6 10 0.55 0.303 -1 1 -0.55
x = 3.05 y =11 = 2.41 = 20 = 4.6
1 2 1
Cov X,X xi x 2.41 0.803
N 1 3
1 1 1
Cov X,Y xi x yi y Cov Y,X yi y xi x 4.6 1.533 and
N 1 N 1 3
1 2 1
Cov Y,Y yi y 20 6.66
N 1 3
Cov X,X Cov X,Y 0.803 1.533
The covariance matrix is A
Cov Y,X Cov Y,Y 1.533 6.666
Cov X,X Var X 0.803 and Cov Y,Y Var Y 6.666
S.D. of X is σ x 0.803 0.896 and S.D. of Y is σ y 6.666 2.582
Cov X,Y 1.15
Correlation coefficient is r 0.497
σ xσ y 0.896 2.582
DSSM Compiled by: Magisetty Harikrushna 3
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Solution: Given data
X 98 87 90 85 95 75
Y 15 12 10 10 16 7
Here number of features n = 2
Number of samples N = 6
1 1
Computation table: Mean values x = ( x i ) and y = ( yi )
N N
x y (xi − x) ( x i − x )2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
98 15 9.667 93.444 3.333 11.111 32.222
87 12 -1.333 1.778 0.333 0.111 -0.444
90 10 1.667 2.778 -1.667 2.778 -2.778
85 10 -3.333 11.111 -1.667 2.778 5.556
95 16 6.667 44.444 4.333 18.778 28.889
75 7 -13.333 177.778 -4.667 21.778 62.222
x =88.333 y =11.667 = 331.333 = 57.333 =125.666
1 2 1
Cov X,X xi x 331.333 66.267
N 1 5
1 1 1
Cov X,Y xi x yi y Cov Y,X yi y xi x 125.666 25.133
N 1 N 1 5
1 2 1
and Cov Y,Y yi y 57.333 11.467
N 1 5
Cov X,X Cov X,Y 66.267 25.133
The covariance matrix is A
Cov Y,X Cov Y,Y 25.133 11.467
Cov X,X Var X 66.267 and Cov Y,Y Var Y 11.467
S.D. of X is σ x 66.267 8.140 and S.D. of Y is σ y 11.467 3.386
Cov X,Y 25.133
Correlation coefficient is r 0.912
σ xσ y 8.140 3.386
4 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Solution: Given data
X 13 15 17 18 19
Y 10 11 12 14 16
Here number of features n = 2
Number of samples N = 5
1 1
Computation table: Mean values x = ( x i ) and y = ( yi )
N N
x y (xi − x) ( x i − x )2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
13 10 -3.400 11.560 -2.600 6.760 8.840
15 11 -1.400 1.960 -1.600 2.560 2.240
17 12 0.600 0.360 -0.600 0.360 -0.360
18 14 1.600 2.560 1.400 1.960 2.240
19 16 2.600 6.760 3.400 11.560 8.840
x =16.4 y =12.6 =23.2 =23.2 = 21.8
1 2 1
Cov X,X xi x 23.2 5.8
N 1 4
1 1 1
Cov X,Y xi x yi y Cov Y,X yi y xi x 21.8 5.45 and
N 1 N 1 4
1 2 1
Cov Y,Y yi y 23.2 5.8
N 1 4
Cov X,X Cov X,Y 5.8 5.45
The covariance matrix is A
Cov Y,X Cov Y,Y 5.45 5.8
Cov X,X Var X 5.8 and Cov Y,Y Var Y 5.8
S.D. of X is σ x 5.8 2.408 and S.D. of Y is σ y 5.8 2.408
Cov X,Y 5.45
Correlation coefficient is r 0.939
σ xσ y 2.408 2.408
X 65.21 64.75 65.56 66.45 65.34
DSSM Compiled by: Magisetty Harikrushna 5
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Y 67.15 66.29 66.2 64.7 66.54 Solution: Given data
Here number of features n = 2
Number of samples N = 5
1 1
Computation table: Mean values x = ( x i ) and y = ( yi )
N N
x y ( x i − x ) ( x i − x )2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
65.21 67.15 -0.252 0.064 0.974 0.949 -0.245
64.75 66.29 -0.712 0.507 0.114 0.013 -0.081
65.56 66.2 0.098 0.010 0.024 0.001 0.002
66.45 64.7 0.988 0.976 -1.476 2.179 -1.458
65.34 66.54 -0.122 0.015 0.364 0.132 -0.044
x = 65.462 y = 66.176 =1.571 =3.273 = −1.827
1 2 1
Cov X,X xi x 1.571 0.393
N 1 4
1 1 1
Cov X,Y xi x yi y Cov Y,X yi y xi x 1.827 0.457
N 1 N 1 4
1 2 1
and Cov Y,Y yi y 3.273 0.818
N 1 4
Cov X,X Cov X,Y 0.393 0.457
The covariance matrix is A
Cov Y,X Cov Y,Y 0.457 0.818
Cov X,X Var X 0.393 and Cov Y,Y Var Y 0.818
S.D. of X is σ x 0.393 0.627 and S.D. of Y is σ y 0.818 0.904
Cov X,Y 0.457
Correlation coefficient is r 0.806
σ xσ y 0.627 0.904
6 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
The Mathematical steps involved in PCA Algorithm are as follows
Step-01: Get data.
1 1
Step-02: Compute the mean vector (µ), using the formulae X x i and Y yi
N N
Step-03: Subtract mean from the given data, to find xi x and yi y
Step-04: Calculate the covariance matrix, using the formulae
1 2 1
Cov X,X xi x , Cov X,Y xi x yi y ,
N 1 N 1
1 1 2
Cov Y,X y i y xi x and Cov Y,Y yi y
N 1 N 1
Cov X,X Cov X,Y
The covariance matrix is
Cov Y,X Cov Y,Y
Step-05: Calculate the Eigen vectors and Eigen values of the covariance matrix.
Step-06: Choose principal component and forming a feature vector.
xj x
Step-07: Derive the new data set, Pij eiT
yj y
DSSM Compiled by: Magisetty Harikrushna 7
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Use PCA to reduce the dimension of the following data
x 2 3 4 5 6 7
y 1 5 3 6 7 8
Solution: Given data
Here number of features n = 2
Number of samples N = 6
1 1
Computation table: Mean of X is x = ( x i ) and Mean of Y is y = ( yi )
N N
x y (xi − x) ( x i − x )2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
2 1 -2.5 6.25 -4 16 10
3 5 -1.5 2.25 0 0 0
4 3 -0.5 0.25 -2 4 1
5 6 0.5 0.25 1 1 0.5
6 7 1.5 2.25 2 4 3
7 8 2.5 6.25 3 9 7.5
x = 4.5 y =5 =17.5 = 34 = 22
1 2 1
Cov X,X xi x 17.5 3.5
N 1 5
1 1 1
Cov X,Y xi x yi y Cov Y,X yi y xi x 22 4.4 and
N 1 N 1 5
1 2 1
Cov Y,Y yi y 34 6.8
N 1 5
Cov X,X Cov X,Y 3.5 4.4
The covariance matrix is A
Cov Y,X Cov Y,Y 4.4 6.8
The characteristic equation is A −I = 0
3.5 λ 4.4
A λI 0
4.4 6.8 λ
The characteristic equation is 2 −10.3+ 4.44 = 0
8 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
The characteristic values are 1 = 0.45; 2 = 9.85
Clearly, the first Eigen value is very small compared to the second Eigen value.
So, the first Eigen value can be left out.
The Eigen vector corresponding to Eigen value = 9.85
6.35 4.4 x
A λI X 0
4.4 3.05 y
6.35x 4.4y 0 (1)
4.4x 3.05y 0 (2)
From equation (2)
x y
3.05 4.4
x = 3.05; y = 4.4
3.05
The Eigen vector is X = , the principal component.
4.4
X 1 3.05 0.57
Now the normalized Eigen vector is e1 = = =
X 5.35 4.4 0.822
xi x
The new data set is, Pij e iT
yi y
2 4.5 2.5
P11 e1T [0.57 0.822] 4.71
1 5 4
3 4.5 1.5
P12 e1T [0.57 0.822] 0.86
3 5 0
4 4.5 0.5
P13 e1T [0.57 0.822] 1.93
3 5 2
DSSM Compiled by: Magisetty Harikrushna 9
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
5 4.5 0.5
P14 e1T [0.57 0.822] 1.11
6 5 1
6 4.5 1.5
P15 e1T [0.57 0.822] 2.45
7 5 2
7 4.5 2.5
P16 e1T [0.57 0.822] 3.89
8 5 3
The new data set is
P11 P12 P13 P14 P15 P16
PCA
-4.71 -0.86 -1.93 1.11 2.45 3.89
10 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Use PCA to reduce the dimension of the following data
X 2.5 0.5 2.2 1.9 3.1 2.3 2.0 1.0 1.5 1.1
Y 2.4 0.7 2.9 2.2 3.0 2.7 1.6 1.1 1.6 0.9
Solution: Given data
Here number of features n = 2
Number of samples N =10
1 1
Computation table: Mean of X is x = ( x i ) and Mean of Y is y = ( yi )
N N
x Y ( x i − x ) ( x i − x ) 2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
2.5 2.4 0.69 0.48 0.49 0.24 0.34
0.5 0.7 -1.31 1.72 -1.21 1.46 1.59
2.2 2.9 0.39 0.15 0.99 0.98 0.39
1.9 2.2 0.09 0.01 0.29 0.08 0.03
3.1 3 1.29 1.66 1.09 1.19 1.41
2.3 2.7 0.49 0.24 0.79 0.62 0.39
2 1.6 0.19 0.04 -0.31 0.10 -0.06
1 1.1 -0.81 0.66 -0.81 0.66 0.66
1.5 1.6 -0.31 0.10 -0.31 0.10 0.10
1.1 0.9 -0.71 0.50 -1.01 1.02 0.72
x =1.81 y =1.91 = 5.55 = 6.45 = 5.54
1 2 1
Cov X,X xi x 5.55 0.616
N 1 9
1 1 1
Cov X,Y xi x yi y Cov Y,X yi y xi x 5.54 0.615 and
N 1 N 1 9
1 2 1
Cov Y,Y yi y 6.45 0.716
N 1 9
Cov X,X Cov X,Y 0.616 0.615
The covariance matrix is A
Cov Y,X Cov Y,Y 0.615 0.716
DSSM Compiled by: Magisetty Harikrushna 11
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The characteristic equation is A −I = 0
0.616 λ 0.615
A λI 0
0.615 0.716 λ
The characteristic equation is 2 −1.332+ 0.063 = 0
The characteristic values are 1 = 0.049; 2 =1.283
Clearly, the first Eigen value is very small compared to the second Eigen value.
So, the first Eigen value can be left out.
The Eigen vector corresponding to Eigen value =1.283
0.667 0.615 x
A λI X 0
0.615 0.567 y
x y
0.567 0.615
Implies x = 0.567; y = 0.615
0.567
The Eigen vector is X = , the principal component.
0.615
X 1 0.567 0.678
Now the normalized Eigen vector is e1 = = =
X 0.836 0.615 0.735
xi x
The new data set is, Pij e iT
yi y
2.5 1.81 0.69
P11 e1T [0.678 0.735] 0.828
2.4 1.91 0.49
0.5 1.81 1.31
P12 e1T [0.678 0.735] 1.778
0.7 1.91 1.21
2.2 1.81 0.39
P13 e1T [0.678 0.735] 0.992
2.9 1.91 0.99
12 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1.9 1.81 0.09
P14 e1T [0.678 0.735] 0.274
2.2 1.91 0.29
3.1 1.81 1.29
P15 e1T [0.678 0.735] 1.676
3 1.91 1.09
2.3 1.81 0.49
P16 e1T [0.678 0.735] 0.913
2.7 1.91 0.79
2 1.81 0.19
P17 e1T [0.678 0.735] 0.099
1.6 1.91 0.31
1 1.81 0.81
P18 e1T [0.678 0.735] 1.145
1.1 1.91 0.81
1.5 1.81 0.31
P19 e1T [0.678 0.735] 0.438
1.6 1.91 0.31
1.1 1.81 0.71
P10 e1T [0.678 0.735] 1.224
0.9 1.91 1.01
The new data set is
P11 P12 P13 P14 P15 P16 P17 P18 P19 P10
PCA
0.828 -1.778 0.992 0.274 1.676 0.913 -0.099 -1.145 -0.438 -1.224
DSSM Compiled by: Magisetty Harikrushna 13
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Use PCA to reduce the dimension of the following data
x 1 2 4 6 2
y 4 8 6 5 7
Solution: Here number of features n = 2
Number of samples N = 5
1 1
Computation table: Mean of X is x = ( x i ) = 3 and Mean of Y is y = ( yi ) = 6
N N
x Y ( x i − x ) ( x i − x ) 2 ( yi − y ) ( y i − y ) 2 ( xi − x)( yi − y)
1 4 -2 4 -2 4 4
2 8 -1 1 2 4 -2
4 6 1 1 0 0 0
6 5 3 9 -1 1 -3
2 7 -1 1 1 1 -1
x =15 y = 30 =16 =10 = −2
1 2 1
Cov X,X xi x 16 4
N 1 4
1 1
Cov X,Y Cov Y,X xi x yi y 2 0.5 and
N 1 4
1 2 1
Cov Y,Y yi y 10 2.5
N 1 4
Cov X,X Cov X,Y 4 0.5
The covariance matrix is A
Cov Y,X Cov Y,Y 0.5 2.5
The characteristic equation is A −I = 0
4 λ 0.5
A λI 0
0.5 2.5 λ
The characteristic equation is 2 − (Trace)+ detA = 0
14 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
i.e., 2 − 6.5+ 9.75 = 0
The characteristic values are 1 = 2.3486; 2 = 4.1514
Clearly, the first Eigen value is small compared to the second Eigen value.
So, the first Eigen value can be left out.
The Eigen vector corresponding to Eigen value = 4.1514
0.1514 0.5 x
Consider A λI X 0
0.5 1.6514 y
x y
1.6514 0.5
Implies x = −1.6514; y = 0.5
−1.6514
The Eigen vector is X = , the principal component.
0.5
X 1 −1.6514 −0.8554
Now the normalized Eigen vector is e1 = = =
X 1.9306 0.5 0.2589
xi x
The new data set is, Pij e iT
yi y
2 2
P11 e1T [ 0.8554 0.2589] 1.193
2 2
1 1
P12 e1T [ 0.8554 0.2589] 1.3732
2 2
1 1
P13 e1T [ 0.8554 0.2589] 0.8554
0 0
3 3
P14 e1T [ 0.8554 0.2589] 2.8251
1 1
1 1
P15 e1T [ 0.8554 0.2589] 1.1143
1 1
DSSM Compiled by: Magisetty Harikrushna 15
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The new data set is
P11 P12 P13 P14 P15
PCA
1.193 1.3732 -0.8554 -2.8251 1.1143
Problem: Use PCA to reduce the dimension of the following data
01. 02.
x 4 8 13 7 x 1 2 3 4
y 11 4 5 14 y 5 7 9 13
16 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Singular Value Decomposition of a Matrix
Mathematically, the singular value decomposition of a matrix can be explained as follows:
Consider a matrix A of order mn .
This can be uniquely decomposed as UVT
Where is mn diagonal matrix, where non-negative real values are called singular values,
1 0 ... 0
i.e., = 0 2 ... 0
0 0 n ...0
V is n n matrix, whose columns are Normalized Eigen Vectors of AT A.
U is m m matrix, whose columns are Normalized Eigen Vectors of AA T or
or
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
DSSM Compiled by: Magisetty Harikrushna 17
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
0 −2
Problem: Obtain the singular value decomposition (SVD) of A, UVT , where A =
1 5
0 −2
Solution: The given matrix is A = of order 2 2
1 5
This can be uniquely decomposed as A = UVT
V is n n matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
0 10 −2
AT A =
−2 51 5
1 5
B = AT A =
5 29
The characteristic equation of B is B −I = 0
1− 5
B−I = =0
5 29 −
2 −30+ 4 = 0
The Eigen values are 1 = 29.866068; 2 = 0.133931
The Eigen vectors for = 29.866068
Consider B −I X = 0
−28.866068 5 x
B−I X = =0
5 −0.866068 y
x −y
=
−0.866068 5
Implies x = 0.866068; y = 5
18 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
x 0.866068
The Eigen vector is X1 = =
y 5
X1 1 0.866068 0.170672
By normalizing we get v1 = = = 0.985327
X1 5.074453 5
The Eigen vectors for = 0.133931
Consider B −I X = 0
0.866069 5 x
B−I X = =0
5 28.866069 y
x −y
=
28.866069 5
Implies x = 28.866069; y = −5
x 28.866069
The Eigen vector is X2 = =
y −5
X2 1 28.866069 0.985327
By normalizing we get v2 = = = −0.170672
X 2 29.295903 −5
0.170672 0.985327
The matrix V = v1 v2 =
0.985327 −0.170672
0.170672 0.985327
VT =
0.985327 −0.170672
1
Now u1 = Av1 here 1 = 1 = 29.866068 = 5.464985
1
1 0 −20.170672
u1 =
5.464985 1 5
0.985327
1 −1.970654
u1 =
5.464985 5.097307
−0.360596
u1 =
0.932721
DSSM Compiled by: Magisetty Harikrushna 19
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
u2 = Av2 , here, 2 = 2 = 0.133931 = 0.365965
2
1 0 −2 0.985327
u2 =
0.365965 1 5
−0.170672
1 0.341344
u2 =
0.365965 0.131967
0.932723
u2 =
0.360600
−0.360596 0.932723
The matrix U = u1 u 2 =
0.932721 0.360600
l 0
The diagonal matrix D ==
0 1
5.464985 0
D ==
0 0.365965
Therefore, singular value decomposition of A is UVT
−0.360596 0.9327235.464985 0 0.170672 0.985327
i.e., A =
0.932721 0.360600 0 0.365965
0.985327 −0.170672
20 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 2
Problem: Obtain the singular value decomposition (SVD) of A, UVT , where A =
2 1
1 2
Solution: The given matrix is A = of order 2 2
2 1
This can be uniquely decomposed as A = UVT
V is 2 2 matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
1 21 2
AT A = B =
2 12 1
5 4
B = AT A =
4 5
The characteristic equation of B is B −I = 0
5 − 4
B−I = =0
4 5 −
2 − (Traceof )+ (detB) = 0
2 − (10)+ (9) = 0
2 −10+ 9 = 0
The Eigen values are 1 = 9 , 2 =1
Eigen vector corresponding 1 = 9
−4 4 x
(B−I)X = = 0
4 −4y
DSSM Compiled by: Magisetty Harikrushna 21
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
x −y
=
−4 4
x y
=
1 1
1
The Eigen vector is X1 =
1
1
X 2
By normalizing v1 = 1 =
X1 1
2
Eigen vector corresponding =1
4 4x
(B−I)X = = 0
4 4y
x −y
=
4 4
x y
Implies =
1 −1
1
The Eigen Vector is X 2 =
−1
1
X 2
By Normalizing v 2 = 2 =
X 2 −1
2
The matrix V = v1 v2
1 1 1 1
2 2 2 2
V= implies V T =
1 −1 1 −1
2 2 2 2
22 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
Now u1 = Av1 here 1 = 1 = 9 = 3
1
1
1 1 2 2
u1 =
3 2 1 1
2
3
1 2
u1 =
3 3
2
1
2
u1 =
1
2
1
u2 = Av2 here 2 = 2 = 1 =1
2
1
1 1 2 2
u2 =
1 2 1 −1
2
−1
2
u2 =
1
2
1 −1
2 2
The matrix U = u1 u 2 = and
1 1
2 2
1 0
The diagonal matrix D = =
0 2
3 0
D ==
0 1
DSSM Compiled by: Magisetty Harikrushna 23
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Therefore, singular value decomposition of A is UVT
1 −1 1 1
2
2 3 0 2 2
i.e., A =
1 1 0 1 1 −1
2 2 2 2
24 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Obtain the singular value decomposition (SVD) of A, UVT , where
1 0 1
A=
−1 1 0
1 0 1
Solution: The given matrix is A = of order 23
−1 1 0
This can be uniquely decomposed as A = UVT
V is 33 matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
1 −1
1 1
AT A = 0 1
0
−1 0
0
1
1
2 −1 1
B = A A = −1 1 0
T
1 0 1
The characteristic equation of B is B −I = 0
2 − −1 1
B −I = −1 1− 0 = 0
1 0 1−
Or
3 − (Traceof A) 2 + (Sumof minorsof PrincipalDiagonal)− (detB) = 0
3 − ( 4) 2 + (3)− (0) = 0
3 − 42 + 3= 0
The Eigen values are 0,1,3
DSSM Compiled by: Magisetty Harikrushna 25
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Eigen vector corresponding = 3
−1 −1 1 x
(B−I)X = −1 −2 0
y = 0
1 0 −2
z
x −y z
= =
4−0 2−0 0+ 2
x y z
= =
4 −2 2
x y z
Implies = =
2 −1 1
2
The Eigen vector is X1 = −1 implies X 1 = ( 2 ) + ( −1) + (1) = 6
2 2 2
1
2
6
X −1
By normalizing v1 = 1 =
X1
6
1
6
Eigen vector corresponding =1
1 −1 1x
(B−I)X = −1 0 0
y = 0
1 0 0
z
x −y z
Expansion by second row = =
0 − 0 0 −1 0 +1
x y z
Implies = =
0 1 1
26 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0
The Eigen vector is X 2 = 1 , implies X 2 = ( 0 ) + (1) + (1) = 2
2 2 2
1
0
X2 1
By normalizing v 2 = =
X2 2
1
2
Eigen vector corresponding = 0
2 −1 1x
(B−I)X = −1 1 0
y = 0
1 0 1
z
x −y z
= =
1− 0 −1+ 0 0 −1
x y z
Implies = =
1 1 −1
1
The Eigen vector is X3 = 1 , implies X 3 = (1) + (1) + ( −1) = 3
2 2 2
−1
1
3
X 1
By normalizing v3 = 3 =
X3
3
−1
3
The matrix V = v1 v2 v3
DSSM Compiled by: Magisetty Harikrushna 27
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
2 1 2 −1 1
0 6 6 6
6 3
−1 1 1 T 1 1
V= implies V = 0
6 2 3 2 2
1 1 −1 1 1 −1
6 3
2 3 3 3
1
Now u1 = Av1 here 1 = 1 = 3
1
2
6
1 1 0 1 −1
u1 =
3 −1 1 0 6
1
6
3
1 6
u1 =
3 −3
6
1
2
u1 =
−1
2
1
u2 = Av2 here 2 = 2 = 1 =1
2
0
1 1 0 1 1
u2 =
1 −1 1 0 2
1
2
28 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
2
u2 =
1
2
1 1
2 2
The matrix U = u1
u2 = and
−1 1
2 2
1 0 0
The diagonal matrix D ==
0 2 0
3 0 0
D = =
0 1 0
Therefore, singular value decomposition of A is UVT
2 −1 1
1 1 6 6 6
2
2 3 0 0 1 1
i.e., A = 0
−1 1 0 1 0 2 2
2 2 1 1 −1
3
3 3
DSSM Compiled by: Magisetty Harikrushna 29
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1 1 0
Problem: Obtain the singular value decomposition (SVD) of A, UVT , where A =
0 1 1
1 1 0
Solution: The given matrix is A = of order 23
0 1 1
This can be uniquely decomposed as A = UVT
V is 33 matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
1 0
1 1 0
A A = 1 1
T
0 1 1
0 1
1 1 0
B = A A = 1 2 1
T
0 1 1
The characteristic equation of B is B −I = 0
2 1 1 0 1 1
+ +
1 1 0 1 1 2
1− 1 0
B −I = 1 2 − 1 = 0
0 1 1−
3 − (Traceof A) 2 + (Sumof minorsof PrincipalDiagonal)− (detB) = 0
3 − ( 4) 2 + (3)− (0) = 0
3 − 42 + 3= 0
The Eigen values are 3,1,0
30 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Eigen vector corresponding = 3
−2 1 0 x
(B−I)X = 1 −1 1
y = 0
0 1 −2
z
x −y z
= =
1 −2 1
x y z
Implies = =
1 2 1
1
The Eigen vector is X1 = 2
1
1
6
X1 2
By normalizing v1 = =
X1 6
1
6
Eigen vector corresponding =1
0 1 0x
(B−I)X = 1 1 1
y = 0
0 1 0
z
x −y z
= =
−1 0 1
x y z
Implies = =
−1 0 1
−1
The Eigen vector is X 2 = 0
1
DSSM Compiled by: Magisetty Harikrushna 31
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
−1
2
X2
By normalizing v2 = = 0
X2
1
2
Eigen vector corresponding = 0
1 1 0x
(B−I)X = 1 2 1
y = 0
0 1 1
z
x −y z
= =
1 1 1
x y z
Implies = =
1 −1 1
1
The Eigen vector is X3 = −1
1
1
3
X −1
By normalizing v3 = 3 =
X3
3
1
3
The matrix V = v1 v2 v3
1 −1 1 1 2 1
6 6 6
6 2 3
2 −1 T −1 1
V= 0 implies V = 0
6 3 2 2
1 1 1 1 −1 1
3 3
6 2 3 3
32 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
Now u1 = Av1 here 1 = 1 = 3
1
1
6
1 1 1 0 2
u1 =
3 0 1 1 6
1
6
3
1 6
u1 =
3 3
6
1
2
u1 =
1
2
1
u2 = Av2 here 2 = 2 = 1 =1
2
−1
2
1 1 1 0
u2 = 0
1 0 1 1
1
2
−1
2
u2 =
1
2
1 −1
2 2
The matrix U = u1 u 2 = and
1 1
2 2
DSSM Compiled by: Magisetty Harikrushna 33
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1 0 0
The diagonal matrix D ==
0 2 0
3 0 0
D = =
0 1 0
Therefore, singular value decomposition of A is UVT
1 2 1
1 −1 6 6 6
2 2 3 0 0 −1 1
i.e., A = 0
1 1 0 1 0 2 2
2 2 1 −1 1
3 3
3
34 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
3 1 1
Problem: Obtain the singular value decomposition (SVD) of A, UVT , where A =
−1 3 1
3 1 1
Solution: The given matrix is A =
− of order 23
1 3 1
This can be uniquely decomposed as A = UVT
V is 33 matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
3 −1
3 1
A A = 1 3
T 1
−1 1
1
3
1
10 0 2
B = A A = 0 10 4
T
2 4 2
The characteristic equation of B is B −I = 0
10 4 10 2 10 0
+ +
4 2 2 2 0 10
10 − 0 2
B −I = 0 10 − 4 =0
2 4 2 −
3 − (Traceof A) 2 + (Sumof minorsof PrincipalDiagonal)− (detA ) = 0
3 − ( 22) 2 + (120)− (0) = 0
3 − 222 +120= 0
The Eigen values are 12,10,0
DSSM Compiled by: Magisetty Harikrushna 35
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Eigen vector corresponding =12
−2 0 2 x
(B−I)X = 0 −2 4
y = 0
2 4 −10
z
x −y z
= =
4 −8 4
x y z
Implies = =
1 2 1
1
The Eigen vector is X1 = 2
1
1
6
X1 2
By normalizing v1 = =
X1 6
1
6
Eigen vector corresponding =10
0 0 2 x
(B−I)X = 0 0 4
y = 0
2 4 −8
z
x −y z
= =
−16 −8 0
x y z
Implies = =
−2 1 0
−2
The Eigen vector is X 2 = 1
0
36 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
−2
5
X2 1
By normalizing v 2 = =
X2 5
0
Eigen vector corresponding = 0
10 0 2x
(B−I)X = 0 10 4
y = 0
2 4 2
z
x −y z
= =
4 −8 −20
x y z
Implies = =
1 2 −5
1
The Eigen vector is X3 = 2
−5
1
30
X3 2
By normalizing v3 = =
X3 30
−5
30
The matrix V = v1 v2 v3
1 −2 1 1 2 1
6 5 30 6 6 6
2 1 2 T −2 1
V= implies V = 0
6 5 30 5 5
1 −5 1 2 −5
0
6 30
30 30 30
DSSM Compiled by: Magisetty Harikrushna 37
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
Now u1 = Av1 here 1 = 1 = 12 = 2 3
1
1
6
1 3 1 1 2
u1 =
2 3 −1 3 1 6
1
6
6
1 6
u1 =
2 3 6
6
1
2
u1 =
1
2
1
u2 = Av2 here 2 = 2 = 10
2
−2
5
1 3 1 1 1
u2 =
10 −1 3 1 5
0
−1
2
u2 =
1
2
1 −1
2 2
The matrix U = u1 u 2 = and
1 1
2 2
38 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 0 0
The diagonal matrix D ==
0 2 0
2 3 0 0
D ==
0 10 0
Therefore, singular value decomposition of A is UVT
1 2 1
−1
1
6 6 6
2 2 2 3 0 0 −2 1
i.e., A = 0
1 1 0 10 0 5 5
2 2 1 2 −5
30
30 30
DSSM Compiled by: Magisetty Harikrushna 39
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1 −1 3
Problem: Obtain the singular value decomposition (SVD) of A, UVT , where A =
3 1 1
1 −1 3
Solution: The given matrix is A = of order 23
3 1 1
This can be uniquely decomposed as A = UVT
V is 33 matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
1 3
1 −1 3
A A = −1 1
T
3 1 1
3 1
10 2 6
B = A A = 2 2 −2
T
6 −2 10
10 − 2 6
The characteristic equation is B −I = 2 2 − −2 = 0
6 −2 10 −
2 −2 10 6 10 2
+ +
−2 10 6 10 2 2
3 − (Trace) 2 + (Sumof monorsof principaldiagonal)− det ( V) = 0
3 − 222 + 96− 0 = 0
3 − 222 + 96= 0
The characteristic values are =16,6,0
The Eigen value corresponding to Eigen value =16
−6 2 6 x
(B−I)X = 2 −14 −2
y = 0
6 −2 −6
z
40 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
x −y z
= =
80 0 80
x y z
Implies = =
1 0 1
1
The Eigen vector X1 = 0
1
1
2
X1
By normalizing v1 = = 0
X1
1
2
The Eigen value corresponding to Eigen value = 6
4 2 6 x
(B−I)X = 2 −4 −2
y = 0
6 −2 4
z
x −y z
= =
−20 20 20
x y z
Implies = =
−1 −1 1
−1
The Eigen vector X 2 = −1
1
−1
3
X −1
By normalizing v 2 = 2 =
X2
3
1
3
The Eigen value corresponding to Eigen value = 0
DSSM Compiled by: Magisetty Harikrushna 41
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
10 2 6 x
(B −I)X = 2 2 −2
y = 0
6 −2 10
z
x −y z
= =
16 32 −16
x y z
Implies = =
−1 2 1
−1
The Eigen vector X3 = 2
1
−1
6
X 2
By normalizing v3 = 3 =
X3
6
1
6
The matrix V = v1 v2 v3
1 −1 −1 1 1
0
2 3 6 2 2
−1 2 −1 −1 1
V= 0 implies V T =
3 6 3 3 3
1 1 1 −1 2 1
6 6
2 3 6 6
1
Now u1 = Av1 here 1 = 1 = 16 = 4
1
1
2
1 1 −1 3
u1 = 0
4 3 1 1
1
2
42 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
4
1 2
u1 =
4 4
2
1
2
u1 =
1
2
1
u2 = Av2 here 1 = 1 = 6
2
−1
3
1 1 −1 3 −1
u2 =
6 3 1 1 3
1
3
3
1 3
u2 =
6 −3
3
1
2
u2 =
−1
2
The matrix U = u1 u 2
1 1
2 2
U=
1 −1
2 2
1 0 0
The diagonal matrix D = =
0 2 0
DSSM Compiled by: Magisetty Harikrushna 43
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
4 0 0
D = =
0 6 0
Therefore, singular value decomposition of A is UVT
1 1
0
1 1
2 2
2 2 4 0 0 −1 −1 1
i.e., A =
1 −1 0 6 0 3 3 3
2 2 −1 2 1
6 6
6
44 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Obtain the singular value decomposition (SVD) of A, UVT ,
3 2 2
where A =
2 3 −2
3 2 2
Solution: The given matrix is A = of order 23
2 3 −2
This can be uniquely decomposed as A = UVT
V is 33 matrix, whose columns are Eigenvectors of AT A.
If V = v1 v2 ... vn implies U = u1 u 2 ... u n
1 1 1
Where u1 = Av1 , u 2 = Av2 … u n = Avn , here 1 = 1 , 2 = 2 … n = n
1 2 n
3 2
3 2
AT A = 2 3
2
−2
−2
2 3
2
13 12 2
B = A A = 12 13 −2
T
2 −2 8
The characteristic equation of V is B −I = 0
13 −2 13 2 13 12
13− 12 2 + +
−2 8 2 8 12 13
B −I = 12 13− −2 = 0
2 −2 8 −
3 − (Traceof A) 2 + (Sumof minorsof PrincipalDiagonal)− (detA ) = 0
3 − (34) 2 + ( 225)− (0) = 0
3 −342 + 225= 0
The Eigen values are 25,9,0
DSSM Compiled by: Magisetty Harikrushna 45
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Eigen vector corresponding = 25
−12 12 2 x
(B−I)X = 12 −12 −2
y = 0
2 −2 −17
z
x −y z
= =
200 −200 0
x y z
Implies = =
1 1 0
1
The Eigen vector is X1 = 1
0
1
2
X1 1
By normalizing v1 = =
X1 2
0
Eigen vector corresponding = 9
4 12 2 x
(B −I)X = 12 4 −2
y = 0
2 −2 −1
z
x −y z
= =
−8 −8 −32
x y z
Implies = =
1 −1 4
1
The Eigen vector is X 2 = −1
4
46 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
3 2
X 2 −1
By normalizing v 2 = =
X2 3 2
4
3 2
Eigen vector corresponding = 0
13 12 2 x
(B −I)X = 12 13 −2
y = 0
2 −2 8
z
x −y z
= =
100 100 −50
x y z
Implies = =
−2 2 1
−2
The Eigen vector is X3 = 2
1
−2
3
X3 2
By normalizing v3 = =
X3 3
1
3
The matrix V = v1 v2 v3
1 1 −2 1 1
0
2 3 2 3 2 2
1 −1 2 T 1 −1 4
V= implies V =
2 3 2 3 3 2 3 2 3 2
0
4 1 −2 2 1
3 2 3 3 3 3
DSSM Compiled by: Magisetty Harikrushna 47
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
Now u1 = Av1 here 1 = 1 = 25 = 5
1
1
2
1 3 2 2 1
u1 =
5 2 3 −2 2
0
5
1 2
u1 =
5 5
2
1
2
u1 =
1
2
1
u2 = Av2 here 2 = 2 = 9 = 3
2
1
3 2
1 3 2 2 −1
u2 =
3 2 3 −2 3 2
4
3 2
3
1 2
u2 =
3 −3
2
1
2
u2 =
−1
2
48 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 1
2 2
The matrix U = u1 u 2 = and
1 −1
2 2
1 0 0
The diagonal matrix D ==
0 2 0
5 0 0
D ==
0 3 0
Therefore, singular value decomposition of A is UVT
1 1
2 0
1 1 2
2 2 5 0 0 1 −1 4
i.e., A =
1 −1 0 3 0 3 2 3 2 3 2
2 2 −2 2 1
3 3 3
DSSM Compiled by: Magisetty Harikrushna 49
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Inner Product of two Vectors
Let u =a1,a 2 ,a 3 ,...a n and v =b1,b2 ,b3 ,...bn be two vectors in R n then the inner product of u
and v is denoted and defined by u,v = a1b1 + a 2 b2 + a 3b3 +...a n bn .
If u =a1,a 2 ,a 3 ,...a n be a vector R n then the norm is defined as u = a12 + a 22 + a 32 +...+ a n2
Orthogonal Vectors
A set of vectors u1,u 2 ,u3 ,...u n are mutually orthogonal if every pair of vectors is orthogonal.
i.e., u i .u j = 0, for all i j .
Orthonormal Vectors
A set of vectors u1,u 2 ,u3 ,...u n is orthonormal if every has magnitude 1 and the set of vectors
are mutually orthogonal.
(i) u i .u j = 0, for all i j (ii) u i =1, for all i
The Gram–Schmidt orthonormalization process
Every finite dimensional inner product space has orthonormal basis.
This method is used to orthonormalizing a given set of vectors in an inner product space.
If u1,u 2 ,u3 ,...u n is a basis for vector space, we can construct a set of orthogonal basis
v ,v ,v ,...v as follows.
1 2 3 n
u1 = v1
u 2 ,v1
v2 = u 2 − 2
v1
v1
u 3 ,v1 u 3 ,v2
v3 = u 3 − 2
v1 − 2
v2 and so on…
v1 v2
Now the orthogonal basis is v1,v2 ,v3 ,...vn
v v v
v
Orthonormal basis is w1 , w 2 , w 3 ,..., w n = 1 , 2 , 3 ,... n
v1 v 2 v3 vn
50 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Given B =u1 ,u 2 ,u 3 , where u1 = (1, 2, 1), u 2 = (1, 1, 3) and u 3 = (2, 1, 1) use the Gram-
Schmidt procedure to Obtain a corresponding orthonormal B.
Solution: Given B =u1 ,u 2 ,u 3 ,
The three vectors are u1 = (1, 2, 1), u 2 = (1, 1, 3) and u 3 = (2, 1, 1)
By the Gram–Schmidt orthonormalization process
u1 = v1 = (1, 2, 1)
Implies v1 = (1) + (2) + (1) = 6
2 2 2 2
u 2 ,v1 = (1, 1, 3).(1, 2, 1) = (1)(1) + (1)(2) + (3)(1) = 6
u 2 ,v1
v2 = u 2 − 2
v1
v1
6
v 2 = (1, 1, 3) − (1, 2, 1)
6
v 2 = (0, −1, 2)
Implies v2 = (0) + (−1) + (2) = 5
2 2 2 2
u3 ,v1 = (2, 1, 1).(1, 2, 1) = (2)(1) + (1)(2) + (1)(1) = 5 and
u3 ,v2 = (2, 1, 1).(0, −1, 2) = (2)(0) + (1)(−1) + (1)(2) =1
u 3 ,v1 u 3 ,v2
v3 = u 3 − 2
v1 − 2
v2
v1 v2
5 1
v3 = ( 2, 1, 1) − (1, 2, 1) − (0, −1, 2)
6 5
7 −7 −7
v3 = , ,
6 15 30
7 −7 −7
v1 = (1, 2, 1) , v 2 = (0, −1, 2) and v3 = , ,
6 15 30
The orthogonal basis is v1, v2 , v3
DSSM Compiled by: Magisetty Harikrushna 51
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
v1 1 1 2 1
w1 = = (1, 2, 1) = , ,
v1 6 6 6 6
v2 1 −1 2
w2 = = (0, −1, 2) = 0, , and
v2 5 5 5
v3 30 7 −7 −7 30 − 30 − 30
w3 = = , , = , ,
v3 7 6 15 30 6 15 30
Orthonormal basis is w1, w 2 , w 3
52 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Given B =u1 ,u 2 ,u 3 , where u1 = (1, 2, 0), u 2 = (8, 1, − 6) and u 3 = (0, 0, 1) use the
Gram-Schmidt procedure to Obtain a corresponding orthonormal B.
Solution: Given B =u1 ,u 2 ,u 3 ,
The three vectors are u1 = (1, 2, 0), u 2 = (8, 1, − 6) and u 3 = (0, 0, 1)
By the Gram–Schmidt orthonormalization process
u1 = v1 = (1, 2, 0)
Implies v1 = (1) + (2) + (0) = 5
2 2 2 2
u 2 ,v1 = (8, 1, −6).(1, 2, 0) = (8)(1) + (1)(2) + (−6)(0) =10
u 2 ,v1
v2 = u 2 − 2
v1
v1
10
v 2 = (8, 1, − 6) − (1, 2, 0)
5
v 2 = (6, − 3, − 6)
Implies v2 = (6) + (−3) + (−6) = 81
2 2 2 2
u3 ,v1 = (0, 0, 1).(1, 2, 0) = (0)(1) + (0)(2) + (1)(0) = 0 and
u3 ,v2 = (0, 0, 1).(6, −3, −6) = (0)(6) + (0)(−3) + (1)(−6) =−6
u 3 ,v1 u 3 ,v2
v3 = u 3 − 2
v1 − 2
v2
v1 v2
0 −6
v3 = (0, 0, 1) − (1, 2, 0) − (6, − 3, − 6)
5 81
4 −2 −4
v3 = , ,
9 9 9
4 −2 −4
v1 = (1, 2, 0) , v 2 = (6, − 3, − 6) and v3 = , ,
9 9 9
The orthogonal basis is v1, v2 , v3
DSSM Compiled by: Magisetty Harikrushna 53
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
v1 1 1 2
w1 = = (1, 2, 0) = , , 0
v1 5 5 5
2 −1 −2
(6, −3, − 6) = , , and
v2 1
w2 = =
v2 9 3 3 3
v3 3 4 −2 −4 2 −1 −2
w3 = = , , = , ,
v3 2 9 9 9 3 3 3
Orthonormal basis is w1, w 2 , w 3
54 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Given u1 = (1, −1, 1), u 2 = (1, 0, 1) and u 3 = (1, 1, 2) use the Gram-Schmidt procedure to
Obtain a corresponding orthonormal basis.
Solution: The three vectors are u1 = (1, −1, 1), u 2 = (1, 0, 1) and u 3 = (1, 1, 2)
By the Gram–Schmidt orthonormalization process
u1 = v1 = (1, −1, 1)
Implies v1 = (1) + (−1) + (1) = 3
2 2 2 2
u 2 ,v1 = (1, 0, 1).(1, −1, 1) = (1)(1) + (0)(−1) + (1)(1) = 2
u 2 ,v1
v2 = u 2 − 2
v1
v1
2
v 2 = (1, 0, 1) − (1, −1, 1)
3
1 2 1
v2 = , ,
3 3 3
2 2 2
2 1 2 1 2
Implies v2 = + + =
3 3 3 3
u3 ,v1 = (1, 1, 2).(1, −1, 1) = (1)(1) + (1)(−1) + (2)(1) = 2 and
1 2 1 1 2 1 5
u3 ,v2 = (1, 1, 2). , , = (1) + (1) + (2) =
3 3 3 3 3 3 3
u 3 ,v1 u 3 ,v2
v3 = u 3 − 2
v1 − 2
v2
v1 v2
2 5 1 2 1
v3 = (1, 1, 2) − (1, −1, 1) − , ,
3 2 3 3 3
−1 1
v3 = , 0,
2 2
1 2 1 −1 1
v1 = (1, −1, 1) , v2 = , , and v3 = , 0,
3 3 3 2 2
DSSM Compiled by: Magisetty Harikrushna 55
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The orthogonal basis is v1, v2 , v3
v1 1 1 −1 1
w1 = = (1, −1, 1) = , ,
v1 3 3 3 3
v2 3 1 2 1 1 2 1
w2 = = , , = , , and
v2 2 3 3 3 6 6 6
v3 −1 1 −1 1
w3 = = 2 , 0, = , 0,
v3 2 2 2 2
Orthonormal basis is w1, w 2 , w 3
56 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Given u1 = (1, 1, 1, 1), u 2 = (1, 1, −1, −1) and u 3 = (0, −1, 2, 1) use the Gram-Schmidt
procedure to Obtain a corresponding orthonormal basis.
Solution: Given three vectors are u1 = (1, 1, 1, 1), u 2 = (1, 1, −1, −1) and u 3 = (0, −1, 2, 1)
By the Gram–Schmidt orthonormalization process
u1 = v1 = (1, 1, 1, 1)
Implies v1 = (1) + (1) + (1) + (1) = 4
2 2 2 2 2
u 2 ,v1 = (1, 1, −1, −1).(1, 1, 1, 1) = (1)(1) + (1)(1) + (−1)(1) + (−1)(1) = 0
u 2 ,v1
v2 = u 2 − 2
v1
v1
0
v 2 = (1, 1, −1, −1) − (1, 1, 1, 1)
4
v 2 = (1, 1, −1, −1)
Implies v2 = (1) + (1) + (−1) + (−1) = 4
2 2 2 2 2
u3 ,v1 = (0, −1, 2, 1).(1, 1, 1, 1) = (0)(1) + (−1)(1) + (2)(1) + (1)(1) = 2 and
u3 ,v2 = (0, −1, 2, 1).(1, 1, −1, −1) = (0)(1) + (−1)(1) + (2)(−1) + (1)(−1) =−4
u 3 ,v1 u 3 ,v2
v3 = u 3 − 2
v1 − 2
v2
v1 v2
2 −4
v3 = (0, −1, 2, 1) − (1, 1, 1, 1) − (1, 1, −1, −1)
4 4
1
v3 = (0, −1, 2, 1) − (1, 1, 1, 1) + (1, 1, −1, −1)
2
1 −1 1 −1
v3 = , , ,
2 2 2 2
1 −1 1 −1
v1 = (1, 1, 1, 1) , v 2 = (1, 1, −1, −1) and v3 = , , ,
2 2 2 2
DSSM Compiled by: Magisetty Harikrushna 57
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The orthogonal basis is v1, v2 , v3
(1, 1, 1, 1) = , , ,
v1 1 1 1 1 1
w1 = =
v1 2 2 2 2 2
1 1 −1 −1
(1, 1, −1, −1) = , , , and
v2 1
w2 = =
v2 2 2 2 2 2
v3 1 1 −1 1 −1 1 −1 1 −1
w3 = = , , , = , , ,
v3 1 2 2 2 2 2 2 2 2
Orthonormal basis is w1, w 2 , w 3
58 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Given u1 = (1, 1, 1, 1), u 2 = (1, 1, 2, 4) and u 3 = (1, 2, − 4, −3) use the Gram-Schmidt
procedure to Obtain a corresponding orthonormal basis.
Solution: Given three vectors are u1 = (1, 1, 1, 1), u 2 = (1, 1, 2, 4) and u 3 = (1, 2, − 4, −3)
By the Gram–Schmidt orthonormalization process
u1 = v1 = (1, 1, 1, 1)
Implies v1 = (1) + (1) + (1) + (1) = 4
2 2 2 2 2
u 2 ,v1 = (1, 1, 2, 4).(1, 1, 1, 1) = (1)(1) + (1)(1) + (2)(1) + (4)(1) =8
u 2 ,v1
v2 = u 2 − 2
v1
v1
8
v 2 = (1, 1, 2, 4) − (1, 1, 1, 1)
4
v 2 = ( −1, −1, 0, 2)
Implies v2 = (−1) + (−1) + (0) + (2) = 6
2 2 2 2 2
u3 ,v1 = (1, 2, − 4, −3).(1, 1, 1, 1) = (1)(1) + (2)(1) + (−4)(1) + (−3)(1) =−4 and
u3 ,v2 = (1, 2, − 4, −3).(−1, −1, 0, 2) = (1)(−1) + (2)(−1) + (−4)(0) + (−3)(2) =−9
u 3 ,v1 u 3 ,v2
v3 = u 3 − 2
v1 − 2
v2
v1 v2
−4 −9
v3 = (1, 2, − 4, − 3) − (1, 1, 1, 1) − (−1, −1, 0, 2)
4 6
3
v3 = (1, 2, − 4, − 3) + (1, 1, 1, 1) + (−1, −1, 0, 2)
2
1 3
v3 = , , −3, 1
2 2
1 3
v1 = (1, 1, 1, 1) , v 2 = ( −1, −1, 0, 2) and v3 = , , −3, 1
2 2
DSSM Compiled by: Magisetty Harikrushna 59
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The orthogonal basis is v1, v2 , v3
(1, 1, 1, 1) = , , ,
v1 1 1 1 1 1
w1 = =
v1 2 2 2 2 2
v2 1 −1 −1 2
w2 = = (−1, −1,0, 2) = , , 0, and
v2 6 6 6 6
v3 21 3 2 3 2 −6 2 2 2
w3 = = , , − 3, 1 = , , ,
v3 5 2 2 10 10 10 10
Orthonormal basis is w1, w 2 , w 3
60 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Regression Line: A regression line is a straight line that describes how a dependent variable
y changes as an independent variable x change. We often use a regression line to predict the
value of y for a given value of x.
Curve fitting by least square method (straight line)
Least Square Method:
The least-squares method is a crucial statistical method that is practiced finding a regression
line or a best-fit line for the given pattern. This method is described by an equation with specific
parameters. The method of least squares is generously used in evaluation and regression. In
regression analysis, this method is said to be a standard approach for the approximation of sets
of equations having more equations than the number of unknowns.
For a given set of ‘n’ data points (x1 ,y1 ),(x 2 ,y 2 ),(x 3 ,y3 )...(x n ,y n ), assume that a
straight line y = f (x) = a + bx → (1) fits the given data in least square sense.
Let the observed value at x xi is y i and the corresponding value on the curve is f (x i )
Error of approximation is d i = yi − f (x i )
n n 2 n 2
Let S = d = yi − f (x i ) = yi − (a + bx i ) →(2)
2
i
i=1 i=1 i=1
We must find the a, b that minimize the function S.
Differentiating (2) partially w.r.to a, b and equating to zero, we get
n 2 n
2
di = yi − (a + bx i ) = 0
a i=1 a i=1
n 2−1
2yi − (a + bx i ) (−1) = 0
i=1
n n n n
y −(a + bx ) = 0 y −a 1− bx = 0
i=1
i i
i=1
i
i=1 i=1
i
n n
yi = na + bx i → (3)
i=1 i=1
DSSM Compiled by: Magisetty Harikrushna 61
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
n 2 n
2
Similarly, di = yi − (a + bx i ) = 0
b i=1 b i=1
n 2−1
2yi − (a + bx i ) (−x i ) = 0
i=1
y − a − bx )(x ) = 0
i=1
i i i
n n n
yi x i = a x i + bx 2i → (4)
i=1 i=1 i=1
Thus, two unknown parameters a and b of equation (1) can be determined by solving the two
equations (3) and (4).
y = na + bx; xy = a x + bx
2
These equations can be written as called as normal
equations.
62 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The following data shows the relationship between the number of hours that 10
persons studied for a French test and their scores on the test.
Hours Studied (x) 4 9 10 14 4 7 12 22 1 17
Test Score (y) 31 58 65 73 37 44 60 91 21 64
Obtain the equation of the least squares line that approximates the regression of the test scores
on the number of hours studied. Also predict the average test score of a person who studied
14 hours for the test.
Solution: Let the required straight line be Y a bX (1)
Here n 10
The two normal equations are Y na b X (2)
XY a X b X 2 (3)
Calculating x, x 2 , y, xy, by the below table
x y x2 xy
4 31 16 124
9 58 81 522
10 65 100 650
14 73 196 1022
4 37 16 148
7 44 49 308
12 60 144 720
22 91 484 2002
1 21 1 21
17 64 289 1088
Σx 100 Σy 544 Σx 2 1376 Σxy 6605
Substituting all the values in normal equations
(2) 10a +100b = 544 → (4)
(3) 100a +1376b = 6605 → (5)
DSSM Compiled by: Magisetty Harikrushna 63
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Solving (4) and (5), we get a = 23.415957, b = 3.098404
Substituting a0 and a1 in equation (1) we get the best fit straight line
Y = 23.415957 +3.098404X
When x 14 we have to find y value.
Y = 23.415957 + 3.098404(14)
Y = 66.793613
When x 14 hours then the score is Y = 67.
64 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Perform the linear regression algorithm for the given data below to predict the
glucose level of the people with age group.
Subject Age (X) Glucose Level (Y)
1 43 99
2 21 65
3 25 79
4 42 75
5 57 87
6 59 81
Solution: Let the required straight line be Y = a + bX → (1)
Here n 6
The two normal equations are Y = na + b X → (2)
XY = a X + b X
2
→ (3)
X , X , Y , XY , by the below table
2
Calculating
X Y X2 XY
43 99 1849 4257
21 65 441 1365
25 79 625 1975
42 75 1764 3150
57 87 3249 4959
59 81 3481 4779
X = 247 Y = 486 X 2 = 11409 XY = 20485
Substituting all the values in normal equations
(2) 6a + 247b = 486 → (4)
(3) 247a + 11409b = 20485 → (5)
Solving (4) and (5), we get a = 65.1416, b = 0.3852
Substituting a0 and a1 in equation (1) we get the best fit straight line Y = 65.1416 + 0.3852 X
DSSM Compiled by: Magisetty Harikrushna 65
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Determine the equation of a straight line which best fits the data.
X 10 12 13 16 17 20 25
Y 10 22 24 27 29 33 37
also find the value of X when Y 25 .
Solution: Let the required straight line be Y a bX
Here n 7
The two normal equations are Y na b X
XY a X b X 2
X Y X2 XY
10 10 100 100
12 22 144 264
13 24 169 312
16 27 256 432
17 29 289 493
20 33 400 660
25 37 625 925
ΣX 113 ΣY 182 ΣX 2 1938 ΣXY 3186
Substituting the values in the normal equations,
Y na b X
XY a X b X 2 , we get
7a 113b 182 (1)
113a 1983b 3168 (2)
By solving these two equations, we get
a 2.63,b 1.45
66 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Hence the equation of the required straight line is Y 2.63 1.45X
When Y 25 ,
25 2.63 1.45X
25 2.63
X 15.427
1.45
When Y 25 , X 15
DSSM Compiled by: Magisetty Harikrushna 67
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Deterministic & Stochastic Statistical Methods
Unit – II
Single Variable Distribution
Problem: Write the distribution for tossing a coin two times.
Let X denotes the number of times head turns up
Sample space of trial is S HH,HT,TH,TT n S 4
Here X is head turns up
Therefore X 0,1,2
1 2 1 1
P X 0 ;P X 1 ;P X 2
4 4 2 4
The distribution is
X 0 1 2
1 1 1
P(x xi )
4 2 4
DSSM Compiled by: Magisetty Harikrushna 1
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A six face fair die is rolled a large number of times. Find the mean value of the outcome.
Let X denotes the random variable.
Sample space of trial is S 1,2,3,4,5,6
The distribution is
X 1 2 3 4 5 6
1 1 1 1 1 1
P(x)
6 6 6 6 6 6
6
Mean μ E(x) xP(x)
x 1
1 1 1 1 1 1
= E ( x ) = 1 + 2 + 3 + 4 + 5 + 6
6 6 6 6 6 6
= 3.5
2 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A discrete random variable X has the following probability distribution
Value of X 1 2 3 4 5 6 7 8
P X x 2k 4k 6k 8k 10k 12k 14k 4k
(i) Find the value of k (ii) Find P X 3 and P X 5 (iii) Find the distribution function of X.
Solution: i) We know that P(xi ) 1
P(x 1) P(x 2) P(x 3) P(x 4) P(x 5) P(x 6) P(x 7) 1
2k 4k 6k 8k 10k 12k 14k 4k 1
60k 1
1
k
60
(ii) Find P X 3 and P X 5
P(x 3) P(x 1) P(x 2)
1 1
P(x 3) 2k 4k 6k 6
60 10
P X 5 P X 5 P X 6 P X 7 P X 8
1 2
P X 5 10k 12k 14k 4k 40k 40
60 3
(iii) Distribution function of X.
Value of X 1 2 3 4 5 6 7 8
1 1 1 2 1 1 7 1
P X x
30 15 10 15 6 5 30 15
1 1 1 1 1 7 14
F x P X x 1
30 10 5 3 2 10 15
DSSM Compiled by: Magisetty Harikrushna 3
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A random variable x has following probability function
X 0 1 2 3 4 5 6 7
P(x xi ) 0 k 2k 2k 3k k2 2k 2 7k 2 k
i) Find k ii) Evaluate P(x 6),P(x 6), P(0 x 5) and P(0 x 4)
1
iii) If P(x k) , find the minimum value of k iv) Mean, Variance and Standard
2
Deviation
Solution: i) We know that P(xi ) 1
P(x 0) P(x 1) P(x 2) P(x 3) P(x 4) P(x 5) P(x 6) P(x 7) 1
0 k 2k 2k 3k k 2 2k 2 7k 2 k 1
10k 2 9k 1 0
k 0.1
ii) P(x 6) P(x 0) P(x 1) P(x 2) P(x 3) P(x 4) P(x 5)
0 k 2k 2k 3k k 2 2k 2 0.8
P(x 6) P(x 6) P(x 7)
2k 2 7k 2 k 0.19
P(0 x 5) P(x 1) P(x 2) P(x 3) P(x 4)
k 2k 2k 3k 0.8
P(0 x 4) P(x 0) P(x 1) P(x 2) P(x 3) P(x 4)
0 k 2k 2k 3k 0.8
1
iii) Given P(x k)
2
1 1
k 0 P(x 0) 0 , It is not True
2 2
4 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 1
k 1 P(x 1) 0 k 0.1 , It is not True
2 2
1 1
k 2 P(x 2) 0 k 2k 0.3 , It is not True
2 2
1 1
k 3 P(x 3) 0 k 2k 2k 0.5 , It is not True
2 2
1 1
k 4 P(x 4) 0 k 2k 2k 3k 0.8 , It is True
2 2
Therefore k 4
7
iv) We know that μ E(x) xi P(xi )
i 0
μ E(x) x0P(x0 ) x1P(x1 ) x 2 P(x 2 ) x 3P(x 3 ) x 4P(x 4 ) x 5P(x 5 ) x 6P(x 6 ) x7 P(x7 )
μ 0 k 4k 6k 12k 5k 2 12k 2 49k 2 7k 3.66
7
Variance σ 2 xi2 P(xi ) μ2 02.0 12.k 2 2.2k 32.2k 42.3k 52.k 2 62.2k 2 3.404
i 0
Standard deviation σ 3.404 1.844
DSSM Compiled by: Magisetty Harikrushna 5
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
k
Problem: Given that f x is a probability distribution for a random variable X that can take on
2x
the values x 0,1, 2, 3, 4. (i) Find k (ii) Mean and variance of X.
k
Solution: Given that f x is a probability distribution of X, where x 0,1, 2, 3, 4.
2x
Here f x is not defined when x 0 , so we take the probability at this point as 0.
4
(i) We Know that f x 1
x 0
f 0 f(1) f(2) f(3) f(4) 1
k k k k
0 1
2 4 6 8
K 1 1 1
1 1
2 2 3 4
24
K
25
Now the probability distribution is
x 1 2 3 4
12 6 4 3
f(x)
25 25 25 25
4
(ii) Mean μ E(x) xf(x)
x 1
μ E(x) 1f(1) 2f(2) 3f(3) 4f(4)
12 6 4 3 48
1 2 3 4
25 25 25 25 25
4 12 2304
Variance σ 2 xf(x) μ2 1 2 3 4
x 1 25 625
6 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
120 2304 3000 2304 696
25 625 625 625
DSSM Compiled by: Magisetty Harikrushna 7
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Two dice are thrown. Let X assign to each point a,b in S the maximum of its number
i.e., X a,b max a,b find the probability distribution. X is a random variable with
X S 1,2,3,4,5,6 . Also find the mean and variance of the distribution.
Solution: Given X assign to each point a,b in S the maximum of its number
i.e., X a,b max a,b
1, 1 , 1, 2 , 1, 3 , 1, 4 , 1, 5 , 1, 6 ,
2, 1 , 2, 2 , 2, 3 , 2, 4 , 2, 5 , 2, 6 ,
3, 1 , 3, 2 , 3, 3 , 3, 4 , 3, 5 , 3, 6 ,
Sample space S
4, 1 , 4, 2 , 4, 3 , 4, 4 , 4, 5 , 4, 6 ,
5, 1 , 5, 2 , 5, 3 , 5, 4 , 5, 5 , 5, 6 ,
6, 1 , 6, 2 , 6, 3 , 6, 4 , 6, 5 , 6, 6
1
Now The number 1 will appear only in once as maximum, so P X 1 1,1
36
3
The number 2 will appear three times as maximum, so P X 2 1,2 , 2,1 , 2,2
36
5
The number 3 will appear five times as maximum, so P X 3
36
7
The number 4 will appear seven times as maximum, so P X 4
36
9
The number 5 will appear nine times as maximum, so P X 5
36
11
The number 6 will appear eleven times as maximum, so P X 6
36
The probability distribution is
x 1 2 3 4 5 6
1 3 5 7 9 11
P(x)
36 36 36 36 36 36
6
Mean μ E(x) xP(x)
x 1
8 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 3 5 7 9 11
μ E(x) 1 2 3 4 5 6
36 36 36 36 36 36
161
4.47
36
Mean μ E(x) 4.47
4 2 2 2 2 2 2
2 2 1 3 5 7 9 11 2
Variance σ xf(x) μ 1 2 3 4 5 6 4.47
x 1 36 36 36 36 36 36
791
Variance σ 2 19.98 1.99
36
x -3 6 9
1 1 1
P(x)
6 2 3
DSSM Compiled by: Magisetty Harikrushna 9
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A variate X has the following probability distribution
Find (i) E(X) (ii) E(X 2 ) (iii) E (2x 1)2 (iv) Mean and variance
Solution: Given
x -3 6 9
1 1 1
P(x)
6 2 3
3 1 1 1 11
(i) E(X) xP(x) 3 6 9
x 0 6 2 3 2
3 2 1 2 1 2 1 93
(ii) E(X 2 ) x2 P(x) 3 6 9
x 0 6 2 3 2
93
E(X 2 )
2
(iii) E (2x 1)2 E (4x 2 4x 1)
4E(X 2 ) 4E(X) 1
93 11
4 4 1 209
2 2
11
(iv) Mean μ E(x)
2
2
variance σ 2 E X 2 E(X)
2
2 93 11 251
σ
2 2 4
10 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Let x denote the sum of the numbers obtained when two fair dice are rolled. Find the
variance and standard deviation of x.
Solution: Let x be the sum of the numbers, this can range from 2 to 12.
1
P (X = 2) = P ({(1, 1)}) =
36
2
P (X = 3) = P ({(1, 2), (2, 1)}) =
36
3
P (X = 4) = P ({(1, 3), (3, 1), (2, 2)}) =
36
4
P (X = 5) = P ({(1, 4), (4, 1), (2, 3), (3, 2}) =
36
5
P (X = 6) = P ({(1, 5), (5, 1), (4, 2), (2, 4), (3, 3)}) =
36
6
P (X = 7) = P ({(1, 6), (6, 1), (2, 5), (5, 2), (3, 4), (4, 3)}) =
36
5
P (X = 8) = P ({(2, 6), (6, 2), (3, 5), (5, 3), (4, 4)}) =
36
4
P (X = 9) = P ({(3, 6), (6, 3), (4, 5), (5, 4)}) =
36
3
P (X = 10) = P ({(4, 6), (6, 4), (5, 5)}) =
36
2
P (X = 11) = P ({(5, 6), (6, 5)}) =
36
1
P (X = 12) = P ({(6, 6)}) =
36
12
Mean μ E(x) xP(x)
x 0
1 2 3 4 1
μ 2. 3. 4. 5. ... 12. 7
36 36 36 36 36
DSSM Compiled by: Magisetty Harikrushna 11
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A sample of 4 items is selected at random from a box containing 12 items of which 5 are
defective. Find the expected number of defective.
Solution: Let x denotes the number of defective items
i.e., x 0,1,2,3,4
Evaluating P(X xi )
5 7
C2 C4 35
P(X 0) 12
0.07,
C4 459
5 7
C1 C3 105
P(X 1) 12
0.35,
C4 459
5 7
C2 C2 210
P(X 2) 12
0.42,
C4 459
5 7
C3 C1 70
P(X 3) 12
0.14,
C4 459
5 7
C4 C0 5
P(X 4) 12
0.01
C4 459
X 0 1 2 3 4
P(X=xi) 0.07 0.35 0.42 0.14 0.01
4
Mean μ E(x) xP(x)
x 0
μ 0. 0.07 1. 0.35 2. 0.42 3. 0.14 4. 0.01 1.65
12 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Take 30 slips of paper and label five each -4 and 4, four each -3 and 3, three each -2 and 2,
and two each -1, 0 and 1. If each slip of paper has the same probability of being drawn find the
probability of getting -4, -3, -2, -1, 0, 1, 2, 3, 4 and find the mean and variance of this distribution.
Solution: Given 30 slips of paper.
Here n s 30 and
The number of – 4 and 4 labelled papers is 5
The number of – 3 and 3 labelled papers is 4
The number of – 2 and 2 labelled papers is 3
The number of – 1, 0 and 1 labelled papers is 2
n E 5 1
Probability of getting – 4 is P 4
n s 30 6
n E 5 1
Probability of getting 4 is P 4
n s 30 6
n E 4 2
Probability of getting – 3 is P 3
n s 30 15
n E 4 2
Probability of getting 3 is P 3
n s 30 15
n E 3 1
Probability of getting – 2 is P 2
n s 30 10
n E 3 1
Probability of getting 2 is P 2
n s 30 10
n E 2 1
Probability of getting – 1 is P 1
n s 30 15
n E 2 1
Probability of getting 0 is P 0
n s 30 15
DSSM Compiled by: Magisetty Harikrushna 13
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
n E 2 1
Probability of getting 1 is P 1
n s 30 15
Therefore, the distribution is
X -4 4 -3 3 -2 2 -1 0 1
1 1 2 2 1 1 1 1 1
P(X=xi)
6 6 15 15 10 10 15 15 15
Mean μ E(x) xP(x)
1 1 2 2 1 1 1 1 1
μ 4 4 3 3 2 2 1 0 1
6 6 15 15 10 10 15 15 15
1 1 2 2 1 1 1 1 1
μ 4 4 3 3 2 2 1 0 1
6 6 15 15 10 10 15 15 15
Mean μ 0
Variance σ 2 x2 f x μ2
21 21 2 2 2 2 2 1 2 1 2 1
= ( −4 ) + ( 4 ) + ( −3) + ( 3) + ( −2 ) + ( −2 ) + ( −1) +
6 6 15 15 10 10 15
( 0 ) + (1) − 02
2 1 2 1
15 15
1 2 1 1
16 16 9 9 4 4 1 1 0
6 15 10 15
32 36 8 2
6 15 10 15
26
Variance σ 2
3
14 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
Problem: Find mean and variance of probability distribution is given by f(x) , for x 1,2,3...n
n
Solution: Given distribution function
X 1 2 3 . . . n
1 1 1 1
P(X=xi) . . .
n n n n
n
Mean: We know that mean μ E(x) xP(x)
x 1
1 1 1 1
μ 1. 2. 3. ... n.
n n n n
1
1 2 3...n
n
1 n(n 1)
n 2
n 1
μ E(x)
2
n
Variance: Variance σ 2 x2 P(x) μ2
x 1
2
1 2 1 2 1 1 n 1
σ 2 12 . 2 . 3 . ... n 2 .
n n n n 2
2
1 2 2 2 2 n 1
1 2 3 ...n
n 2
2
1 n(n 1)(2n 1) n 1
n 6 2
(n 1) (2n 1) (n 1)
2 3 2
DSSM Compiled by: Magisetty Harikrushna 15
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
(n 1) 4n 2 3n 3
2 6
(n 1) n 1
2 6
2 n2 1
σ
12
16 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Continuous type random variable
A random variable which takes the values continuously, i.e., all possible values in a given
interval is called as continuous type random variable.
Eg: 01. The height of a student in a particular class may be between 5 and 8
i.e., X {x:5 x 8}
02. Time and temperature are the continuous random variables.
Probability density function:
Let X be a continuous type of random variable. Let a function f(x) be such that
(i) f x 0, x S (ii) f(x)dx 1
S
Then the function f(x) is called as Probability density function.
Note: 01. If X is a continuous type of random variable with Probability mass function f(x), then
a b
P(x a) f(x)dx 0 , P(a x b) f(x)dx
a a
02. For X a random variable whose distribution function F x has a derivative. The function f t
x d
satisfying F x f(t)dt. Implies Probability density function f x F x .
dx
Mean (Expectation): Let X be a continuous type random variable with respective probabilities
P(x) f(x) , for x , then expectation or Mean is defined as xf(x)dx and is denoted by
E(x) or μ i.e., E(x) μ xf(x)dx
Variance: The variance of continuous distribution is given by σ 2 x2 f(x) μ2
DSSM Compiled by: Magisetty Harikrushna 17
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Median: Median is the point which divides the entire distribution into two equal parts.
In the case of continuous distribution, median is the point which divides the total area into two equal
b
parts. Thus, if X is defined from a to b and M is the median, then f(x)dx 1
a
M b
f(x)dx f(x)dx 1
a M
M 1
f(x)dx (1) and
a 2
b 1
f(x)dx (2)
M 2
by solving equations (1) and (2) we get the median M.
Mode: Mode is the value of x for which f(x) is maximum. Thus, mode is given by f (x) 0 and
f '(x) 0, a x b
18 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
k
Problem: Given the probability density function of random variable X as f(x) , x ,
1 x2
find k.
k
Solution: Given f(x) , x ,
1 x2
We know that f(x)dx 1
1
k dx 1
1 x2
k tan 1 x 1
1 1
k tan tan 1
π π
k 1
2 2
1
k
π
DSSM Compiled by: Magisetty Harikrushna 19
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A probability density function on the interval a,1 is given by x −2 and outside this interval
is the value of the function is zero. Then find the value of a.
1
Solution: Given f(x) ,a x 1,
x2
We know that f(x)dx 1
1 1
i.e., 2
dx 1
ax
1
1
1
x a
1
1 1
a
1
2
a
1
a
2
20 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A continuous random variable X has the distribution function
0 if x 1
4
F(x) k(x 1) if 1 x 3
1 if x 3
Find (i) f(x) (ii) k (iii) Mean (iv) Variance
0if x 1
4
Solution: Given F(x) k(x 1) if 1 x 3
1 if x 3
0 if x 1
d 3
(i) We know that f(x) F(x) 4k(x 1) if 1 x 3
dx
0 if x 3
(ii) We know that f(x)dx 1
1 3
f(x)dx f(x)dx f(x)dx 1
1 3
1 3
(0)dx 4k (x 1)3 dx (0)dx 1
1 3
3
(x 1)4
4k 1
4
1
k 24 0 1
1
k
16
DSSM Compiled by: Magisetty Harikrushna 21
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
(iii) Mean μ E(x) xf(x)dx
1 3
xf(x)dx xf(x)dx xf(x)dx
1 3
1 3
x(0)dx 4k x(x 1)3 dx x(0)dx
1 3
3
4k x(x 1)3 dx
1
3
4k (x 4 3x3 3x2 x)dx
1
3
1 x5 x4 x3 x2
4 3 3
16 5 4 3 2
1
5 5 4 5 2 2
1 3 1 3 1 3 3 3 1
3 3 1
4 5 4 2
μ E(x) 2.6
(iv) Variance σ 2 x2 f(x)dx μ2
1 3
x2 f(x)dx x2 f(x)dx x2 f(x)dx μ2
1 3
1 3
x(0)dx 4k x(x 1)3 dx x(0)dx (2.6)2
1 3
3
4k x2 (x 1)3 dx (2.6)2
1
3
4k (x5 3x4 3x3 x2 )dx (2.6)2
1
22 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
3
1 x6 x 5 x 4 x 3
4 3 3 (2.6)2
16 5 5 4 3
1
6 6 5 5 4 4 3 3 3
1 3 1 3 1 3 1 3 1
4 3 3 6.8 0.1066
4 5 5 4 3
1
Variance σ 2 0.1066
DSSM Compiled by: Magisetty Harikrushna 23
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
x
Problem: The probability density function f(x) ce , x . is a continuous type. Show that
1
c and find mean and variance of the distribution and also find the probability that lies between 0
2
and 4.
x
Solution: Given f(x) ce , x
We know that f(x)dx 1
x 1
ce dx 1 c.2. e x dx 1 2c 1 c
0 2
Mean μ E(x) xf(x)dx
1 x
μ xe dx 0 Since f(x) is odd function
2
Variance σ 2 x2 f(x)dx μ2
1 x
σ2 x2 e dx μ2
2
1 x
.2 x2 e dx
2 0
x2 e x dx
0
x3 1 e x dx
0
Γ 3 By gamma function xn 1 e x dx Γ n
0
2!
24 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Variance σ 2 2
14 x
P(0 x 4) e dx
20
14 x
e dx
20
1 x 4
e
2 0
1 4
1 e
2
DSSM Compiled by: Magisetty Harikrushna 25
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: If x is a continuous random variable, Y aX b prove that E(Y) aE(X) b and
V(Y) a 2 V(X) where V is the variance and a, b are constants.
Solution: Given Y aX b
We know that E(y) yf(x)dx (ax b)f(x)dx
a xf(x) b)dx b f(x)dx aE(x) b
E(y) aE(x) b
2
We know that V(y) E(y 2 ) E(y)
2
V(y) E (ax b)2 E(ax b)
2
V(y) E (a 2 x 2 2abx b 2 ) aE(x) b
2
V(y) a 2 E x 2 2abE x E b2 a 2 E(x) 2abE x E b2
2
V(y) a 2 E x 2 a 2 E(x)
2
V(y) a 2 E x 2 E(x) a 2 V(x)
V(y) a 2 V(x)
26 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
Sinx, 0 x π
Problem: The probability density function of continuous random variable is f(x) 2
0 Other wise
Find (i) Mean (ii) Median (iii) Mode
Solution: i) Mean E(x) μ xf(x)dx
1π
xSinxdx
20
1 π
xcosx sinx 0
2
1
πcosπ sinπ 0 0
2
1
π 1 0
2
π
Mean E(x) μ
2
M π 1
1
ii) Let M be the median of distribution then f(x)dx and f(x)dx
0 2 M 2
1M 1
Sinxdx
20 2
M
Cosx 0
1
π
CosM 0 M
2
iii) Since mode is the value of x for which f(x) is maximum.
1
f(x) Sinx
2
1
f (x) Cosx
2
DSSM Compiled by: Magisetty Harikrushna 27
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
For point of maximum, we equate f (x) to zero.
i.e., f (x) 0
1
Cosx 0
2
π
x
2
1
Now f (x) Sinx
2
π 1 π
f Sin
2 2 2
1
0
2
π
x is the point of maximum.
2
π
Mode of f(x) is
2
28 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The cumulative distribution function for continuous random variable X is
1 e 2x x 0
F(x) find (i) f(x) (ii) Mean (iii) Variance and Standard deviation
0 Otherwise
1 e 2x x 0
Solution: Given cumulative distribution function F(x)
0 Otherwise
2x x 0
d 0 2e
(i) We know that F(x) f x
dx 0 Otherwise
2e 2x x 0
f x
0 Otherwise
(ii) Mean E(x) μ xf(x)dx
0
2x
E(x) μ x 0 dx x 2e dx
0
2x
μ 2 xe dx
0
x 2x 1 2x
μ 2 e e
2 4 0
1
μ 2 0 0 0 1
4
1
μ
2
Variance σ 2 x2 f(x)dx μ2
0
σ2 x2 f(x)dx x2 f(x)dx μ2
0
DSSM Compiled by: Magisetty Harikrushna 29
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
0 2
1
σ2 x2 0 dx x2 2e 2x
dx
0 2
1
σ 2 2 x2 e 2x
dx
0 64
x2 x 1 1
σ2 2 e 2x
e 2x
e 2x
2 2 4 4
0
1 1
σ2 2 0 1
4 4
1 1
σ2
2 4
1
Variance σ 2
4
30 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: For a continuous random variable the probability distribution function
λx x 0
kxe
f(x) find (i) k (ii) Mean (iii) Variance
0 Otherwise
λx x 0
kxe
Solution: Given probability distribution function f(x)
0 Otherwise
(i) We have f(x)dx 1
0
f(x)dx f(x)dx 1
0
λx
0 k xe dx 1
x λx 1 λx
k e 2
e 1
λ λ 0
1
k 0 0 1 1
λ2
k
1
λ2
k λ2
(ii) Mean E(x) μ xf(x)dx
0
λx
E(x) μ x 0 dx x kxe dx
0
μ k x2 e λx
dx
0
2 x2 2x 2x 2x 2 2x
μ λ e 2
e 3
e
λ λ λ 0
DSSM Compiled by: Magisetty Harikrushna 31
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
2
μ λ2 0 1
λ3
2
Mean E(x) μ
λ
Variance σ 2 x2 f(x)dx μ2
0
σ2 x2 f(x)dx x2 kxe λx
dx μ2
0
0 2
2
σ2 x2 0 dx k x3e λx
dx
0 λ
2
2 2 3 λx 2
σ λ x e dx
0 λ
2
x3 3x 2 6x 6 2
σ 2 λ2 e λx
e λx
e λx
e λx
λ λ2 λ3 λ4 0
λ
6 4
σ 2 λ2 0 4
1
λ λ2
6 4
σ2 2
λ λ2
2
Variance σ 2
λ2
Problem: For a continuous random variable the probability distribution function
kxe 2x x 0
f(x) find (i) k (ii) Mean (iii) Variance
0 Otherwise
32 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The daily consumption of electric power (in millions of KW -hours) is a random variable
x
1 3
having the probability density function f(x) xe x 0 If the total production is 12 million
9
0 x 0
KW-hour, determine the probability that there is power cut (shortage) on any given day.
x
1 3
Solution: Given probability density function f(x) xe x 0
9
0 x 0
The probability that the power consumed is between 0 and 12 is P 0 x 12
12
i.e., P 0 x 12 f x dx
0
x
1 12 3
P 0 x 12 xe dx
90
x x 12
1
3xe 3 9e 3
9
0
1 4 4
3 12 e 9e 0 9 1
9
1 4 4
36e 9e 9
9
1 4
45e 9
9
9 4
5e 1
9
4
P 0 x 12 1 5e
Power supply is inadequate if daily consumption exceeds 12 million KW.
i.e., P x 12
DSSM Compiled by: Magisetty Harikrushna 33
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
P x 12 1 P x 12
P x 12 1 P 0 x 12
4
P x 12 1 1 5e
4
P x 12 5e
34 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A continuous random variable x has a probability density function f(x) 3x 2 ,0 x 1, find
a and b such that (i) P(x a) P x a and (ii) P(x b) 0.05 .
Solution: Given f(x) 3x 2 ,0 x 1,
(i) P(x a) P x a
1
P(x a) P x a Since total probability is 1
2
1
P(x a)
2
a 1
f(x)dx
0 2
a 1
3x2 dx
0 2
a
x3 1
3
3 2
0
1
a3 0
2
1
1 3
a
2
(ii) P(x b) 0.05
1 1
f(x)dx
b 20
1 1
3x2 dx
b 20
1
x3 1
3
3 20
b
DSSM Compiled by: Magisetty Harikrushna 35
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
1 b3
20
1
b3 1
20
1
19 3
b
20
0
if x 2
1
Problem: If the function is defined by f(x) (2x 3) if 2 x 4 a probability density function?
18
if x 4
0
find the probability that variate having f x as density function will fall in the interval 2 x 3.
Problem: For the continuous probability function f x kx 2 e x when x 0, find (i) k (ii) Mean
(iii) Variance
Problem: The diameter of an elective cable say X is assumed to be continuous random variable with
p.d.f of f x kx 1 x2 in 0 x 1, zero elsewhere find the value of k and
1 1
P 0 x , P x .
2 4
36 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Probability Distributions
Discrete Theoretical Distributions
Binomial distribution:
Let ‘n’ be a positive integer, let 0 p 1 .
n− x
Let f ( x) = Cx p q , x = 0,1,2...n, zero elsewhere and q = 1 − p p + q = 1
n x
n
Clearly f(x) 0 and f(x) 1
x 0
Therefore f(x) satisfies the conditions of being a Probability distribution function
Definition: Any random variable X whose p.m.f is the above f(x) is said to have a Binomial
distribution and the p.m.f. f(x) is called the binomial p.m.f.
Note: In the Binomial distribution n and p are the parameters.
DSSM Compiled by: Magisetty Harikrushna 37
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
n
Mean of Binomial distribution: Mean μ E(x) xP(x)
x 0
n
μ xP(x)
x 0
n
x n C x px q n x
x 0
n 1 n 1 n 2 n 2 n 3 n 3 n n n n
0 (1) C 1p q (2) C2p q (3) C 3p q ... (n) Cnp q
n(n 1) 2 n n(n 1)(n 2) 3 n
n pq n 1
2 p q 2
3 p q 3
... np n
2! 3!
(n 1)(n 2) 2 n
n p qn 1
(n 1)pq n 2
pq 3
... pn 1
2
n 1 n 2 2 n 3
np (n 1)
C0 q
(n 1)
C1pq
(n 1)
C2p q ... (n 1)
C(n 1) pn 1
n 1
np q p
np(1)n 1
np
Mean μ E(x) np
n
2
Variance of Binomial distribution: Variance σ x2 P(x) μ2
x 0
n
σ2 (x2 x x)P x μ2
x 0
n n
x(x 1)P x xP x μ2
x 0 x 0
n n
x(x 1)P x μ μ2 μ xP x
x 0 x 0
n
0 0 x(x 1) n Cxpxq n x
μ μ2
x 2
38 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
= 2(1) nC2 p2 qn−2 + 3(2) nC3 p3qn−3 + 4(3) nC4 p4 q n−4 + ... + nCn p n + − 2
n(n − 1) 2 n − 2 n( n − 1)(n − 2) 3 n −3 n( n − 1)(n − 2)(n − 3) 4 n − 4 n( n − 1)... n
=2 p q +6 p q + 12 p q + ... + p + − 2
2! 3! 4! n!
(n − 2)(n − 3) 2 n−4 (n − 2)(n − 3)... n−2
= n(n − 1) p2 q n−2 + (n − 2) pq n−3 + p q + .. + . p + − 2
2! n!
= n(n − 1) p2 ( ( n −2)C0 q n−2 + ( n −2)C1 pq n−3 + ( n −2)C2 p 2 q n−4 + ... + ( n −2)Cn −2 p n−2 ) + − 2
n 2
n(n 1)p2 q p μ μ2
n 2
n(n 1)p2 1 μ μ2
(n 2 n)p2 μ μ2
n 2 p2 n p2 np n 2 p2
2 = np (1 − p )
Therefore variance 2 = npq
Standard deviation = npq
Mode of the Binomial Distribution:
Mode of the Binomial distribution is the value of x that maximizes the probability function.
Integralpartof n 1 p, if n 1 p is not an integer
Mode
n 1 p, if n 1 p is an integer
Recurrence relation for Binomial distribution:
n− x x +1
We know that P( x) = Cx p q
n x
and P( x + 1) = Cx +1 p
n
qn− x−1
P( x + 1) nCx +1 p x +1 q n− x −1
= n x n− x
P( x) Cx p q
DSSM Compiled by: Magisetty Harikrushna 39
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
P(x 1) n x p
P(x) x 1 q
n x p
P(x 1) P(x) is known as Recurrence relation for Binomial distribution.
x 1 q
40 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
Problem: The probability that a man hitting target is , if he fires 6 times, find the probability that
3
(i) At most 5 times (ii) Exactly 1 time (iii) At least 2 times
1
Solution: Given that the probability of man is p
3
1 2
q (1 p) 1 , and n 6
3 3
n− x
We know that Binomial distribution function is P( x) = Cx p q
n x
(i) At most 5 times P(X 5) 1 P(X 5) 1 P(X 6) Total Probability is 1
6 6 6
6 1 2 0 1 2 3 4 5 6
1 C6 0.9986
3 3
1 6 1
1 2
(ii) Exactly 1 time P(X 1) 6
C1 0.2633 Total Probability is 1
3 3
0 1 2 3 4 5 6
(iii) At least 2times P(X 2) 1 P(X 2)
1 P(X 0) P(X 1)
0 6 0 1 6 1
6 1 2 6 1 2
C0 C1 0.6488
3 3 3 3
1
Problem: The probability that John hits a target is . He fires 6 times, find the probability that he
2
hits the target (i) Exactly 2 times (ii) More than 4 times (iii) At least once.
1
Problem: The probability that a man hitting target is , if he fires 7 times, find the probability of
4
hitting the target at least twice.
DSSM Compiled by: Magisetty Harikrushna 41
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: 10 coins are thrown simultaneously, find the probability of getting at least
(i) 7 heads (ii) 6 heads.
1
Solution: Let p be the probability of getting a head, i.e., p
2
1
q (1 p) . here n 10.
2
n− x
We know that Binomial distribution function is P( x) = Cx p q
n x
(i) At least 7 heads
P(x 7) P(x 7) P(x 8) P(x 9) P(x 10)
10 x 10 x
10 1 1
P(x 7) Cx
x 7 2 2
10 10
10 1
Cx
x 7 2
10
1 10 10 10 10
C7 C8 C9 C 10
2
11
P(x 7) .
64
(ii) At least 6 heads
P(x 6) P(x 6) P(x 7) P(x 8) P(x 9) P(x 10)
10 x 10 x 10 10 10
10 1 1 10 1 1 10
P(x 6) Cx Cx C6 10 C7 10 C8 10 C9 10 C10
x 6 2 2 x 6 2 2
193
P(x 6) .
512
Problem: The mean and variance of a binomial variable X with parameters 'n' and 'p' are 16 and 8.
Find P X 1 and P X 2 .
42 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Two dice are thrown five times; find the Probability of getting 7 as sum. (i) At least once
(ii) 2 times (iii) P(1 x 5)
Solution: Here, n = 5
Let p be the probability of getting sum 7.
n(E) 6 1
i.e., p
n(S) 36 6
1 5
q 1 p 1
6 6
E {(1,6),(2,5),(3,4),(4,3),(5,2),(6,1)}
n− x
The Binomial distribution function is P( x) = Cx p q
n x
(i) At least once P(x 1) 1 P(x 1) 1 P(x 0) Total Probability is 1
0 5 0
5 1 5 0 1 2 3 4 5
1 C0
6 6
Probability of getting 7 as sum at least once is P(x 1) 0.5981
(ii) 2 times
2 5 2
5 1 5
i.e., P(x 2) C2
6 6
Probability of getting 7 as sum at least once is P(x 2) 0.1607
(iii) P(1 x 5)
i.e. P(x 2) P(x 3) P(x 4)
2 5 2 3 5 3 4 5 4
5 1 5 5 1 5 5 1 5
C2 C3 C4
6 6 6 6 6 6
P(1 x 5) 0.19611
DSSM Compiled by: Magisetty Harikrushna 43
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Out of 800 families with 5 children each, how many would you expect to have (i) 3 boys
(ii) either 2 or 3 boys (iii) 5 girls (iv) At least one boy
Solution: Here the probabilities of boy and girl are equal.
1 1 1
Let p be the probability of boy p implies probability of girl q 1 ,
2 2 2
No. of children in each family is n 5
n− x
The probability p of x successes is given by P( x) = Cx p q
n x
(i) Having 3 boys
3 5 3
5 1 1 5
i.e., P(X 3) C3
2 2 16
5
For 800 families the probability of number of families having 3 boys is 800 250
16
(ii) Either 2 boys or 3 boys P(X 2) P(X 3)
2 5 2 3 5 3
5 1 1 5 1 1
C2 C3
2 2 2 2
5
P(X 2) P(X 3)
8
5
For 800 families the probability of number of families having 3 boys is 800 500
8
(iii) 5 girls is equal to no boy
0 5 0
5 1 1
i.e., P(X 0) C0
2 2
1
P(X 0)
32
1
For 800 families the probability of number of families having 3 boys is 800 25
32
44 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
(iv) At least one boy P(X 1) 1 P(X 1) 1 P(X 0)
0 5 0
5 1 1 1
1 C0 1
2 2 32
31
P(X 1)
32
31
For 800 families the probability of number of families having 3 boys is 800 775
32
Problem: In a family of 5 children find the probability that there are (i) 2 boys (ii) At least one
boy (iii) All are boys (iv) No boys.
DSSM Compiled by: Magisetty Harikrushna 45
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: It has been claimed that in 60% of all solar heat installations the utility bill is reduced by at
least one-third. Accordingly, what are the probabilities that the utility bill will be reduced by at least
one-third in (i) four of five installations (ii) at least four of five installations.
Solution: Given that the probability that in the solar heat installations the utility bill is reduced by
one-third is p 60% 0.6
q 1 p 0.4
Here n 5
x
The Binomial distribution function is P(x)
n
Cxp (1 p)n x
(i) Four of Five Installations
5 4 5 4
i.e., P(x 4) C 4 0.6 0.4
P(x 4) 0.2592
(ii) At least four of five installations
i.e., P(X 4) P(X 4) P(x 5)
5 4 5 4 5 5 5 5
C 4 0.6 0.4 C 5 0.6 0.4
P(X 4) 0.337
46 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: 20% of items are produced from a factory are defective. Find the probability that in a
sample of 5 chosen at random such that (i) None is defective (ii) one is defective
(iii) Find P 1 x 5
20
Solution: Given p 20% 0.2
100
(1 p) 1 0.2 0.8, and n 5.
x
The Binomial distribution function is P(x)
n
Cxp (1 p)n x
(i) None is defective
5 0 5 0
i.e., P(X 0) C 0 0.2 (0.8)
P(X 0) 0.32768
(ii) One is defective
5 1 5 1
i.e., P(X 1) C 1 0.2 (0.8)
P(X 1) 0.4096
(iii) P(1 X 5)
4
x
P(1 X 5) n
Cxp (1 p)n x
x 2
2 3 4
P(1 X 5) 5
C 2 0.2 (0.8)5 2 5
C 3 0.2 (0.8)
5 3 5
C 4 0.2 (0.8)5 4
P(1 X 5) 0.2624
DSSM Compiled by: Magisetty Harikrushna 47
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Assume that 50% of all engineering students are good in Mathematics. Determine the
probabilities that among 18 engineering students
(i) Exactly 10 (ii) At least 10 (iii) At most 8 (iv) At least 2 and at most 9 are good in
Mathematics
1
Solution: Given n 18; p 50%
2
1 1
q 1 p 1
2 2
n x n x
The probability p of x successes is given by P(X x) C xp q
(i) Exactly 10
10 18 10
18 1 1
i.e., P(x 10) C10
2 2
P x 10 0.1670
(ii) At least 10
10
x 18 x 10
18
18 1 1 18 1
i.e., P(x 10) Cx Cx
x 0 2 2 x 0 2
18
1 18
P(x 10) C0 18 C1 18 C 2 18 C 3 18 C 4 18 C 5 18 C6 18 C7 18 C8 18 C9 18 C10
2
(iii) At most 8
8
x 18 x 8
18
18 1 1 18 1
i.e., P(x 8) Cx Cx
x 0 2 2 x 0 2
18
1 18
P(x 8) C0 18 C1 18 C 2 18 C 3 18 C 4 18 C 5 18 C6 18 C7 18 C8
2
48 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A box contains 9 cards numbered 1 to 9. If four cards are drawn with replacement. What is
the probability that none is 1?
Solution: Given n 4
1
Let p be the probability of getting 1 p
9
1 8
q 1 p 1
9 9
n x n x
The probability p of x successes is given by P(X x) C xp q
Therefore, the probability of getting none is 1 is given by
0 4 0
4 1 8
P(X 0) C0 0.624
9 9
P(X 0) 0.624
DSSM Compiled by: Magisetty Harikrushna 49
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A die is thrown 8 times. If getting 2 or 4 is a success. Find the probability of
(i) 4 success (ii) P(X 3) (iii) P(X 2)
Solution: Given n 8,
1 1 1 1 2
Let p be the probability of getting 2 or 4 p q 1 p 1
6 6 3 3 3
n x n x
The probability p of x successes is given by P(X x) C xp q
(i) 4 successes
4 8 4
8 1 2
i.e., P(X 4) C4 0.17
3 3
P(X 4) 0.17
(ii) P(X 3) P(X 0) P(X 1) P(X 2) P(X 3)
0 8 0 1 8 1 2 8 2 3 8 3
8 1 2 8 1 2 8 1 2 8 1 2
C0 C1 C2 C3
3 3 3 3 3 3 3 3
P X 3 0.74
(iii) P(X 2)
i.e., P(X 2) 1 P X 2 1 Px 0 Px 1
0 8 0 1 8 1 2 8 2
8 1 2 8 1 2 8 1 2
1 C0 C1 C2
3 3 3 3 3 3
P(X 2) 0.8049
50 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: If the probability of a defective bolt is 1/8, find
(i) The mean (ii) The variance of the distribution of defective bolts of 640
1 1 7
Solution: Given n 640; p q 1 p 1
8 8 8
640
(i) We know that mean μ np 80
8
7
(ii) Variance = npq = 640 ( 80 ) = 70
2
8
DSSM Compiled by: Magisetty Harikrushna 51
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: The mean and variance of Binomial distribution are 4 and 4/3 find P(X 1)
4
Solution: Given that mean np 4, variance npq
3
4 1 1 2
4q q p 1 and
3 3 3 3
2
np 4 n 4
3
n 6
Now P(X 1) 1 P(X 1) 1 P(X 0)
0 6 0
6 2 1
1 C0
3 3
P(X 1) 0.9986
52 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: In a binomial distribution consisting of 5 independent trials, probability of 1 and 2
successes are 0.4096 and 0.2048 respectively. Find the parameter ‘p’ of the distribution.
Solution: Given n 5;
P(X 1) 0.4096 and P(X 2) 0.2048
i.e., P(X 1)
5
C1p1q 5 1 0.4096
5pq 4 0.4096 pq 4 0.08192 (1)
P(X 2) 5 C 2 p2 q 5 2 0.2048
10p2 q 3 0.2048 p2 q 3 0.02048 (2)
(1) pq 4 0.08192 q
2 3
4
2 p q 0.02048 p
(1 p) 1
4 1 p 4p 5p 1 p
p 5
DSSM Compiled by: Magisetty Harikrushna 53
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Binomial frequency distribution
The possible number of successes and their frequencies is called a Binomial Frequency
distribution. In N sets of n trials the theoretical frequencies of 0, 1, 2, …. n successes will be given by
n
the terms of expansion of N q p
The theoretical frequencies are given by N P x
n x p
Were, P x 1 .P x
x 1 q
54 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Fit a Binomial distribution to the following data
x 0 1 2 3 4 5
f 2 14 20 34 22 8
Solution: Here n 5
Evaluation of Mean
x f xf Probability P x f x N.P x
0 2 0 P x 0 0.015 1.5
1 14 14 P x 1 0.098 9.8
2 20 40 P x 2 0.257 25.7
3 34 102 P x 3 0.337 33.7
4 22 88 P x 4 0.223 22.3
5 8 40 P x 5 0.058 5.8
N Σfi 100 Σfi xi 284
Σfi xi 284
Mean μ 2.84
Σfi 100
2.84
i.e., mean np 2.84 p 0.568
5
q 1 p 1 568 0.432
P x 0 n
C 0pxq n x
5 0 5 0
P x 0 C0 0.568 0.432
5
P x 0 0.432 0.015
n x p
We have recurrence relation P x 1 .P x
x 1 q
DSSM Compiled by: Magisetty Harikrushna 55
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
5 0 0.568 5 1 0.568
P 1 0.015 0.098 , P 2 0.098 0.257
0 1 0.432 1 1 0.432
5 2 0.568 5 3 0.568
P 3 0.257 0.337 , P 4 0.34 0.223
2 1 0.432 3 1 0.432
5 4 0.568
P 5 0.223 0.058
4 1 0.432
The theoretical frequencies are 1.5, 9.89, 26, 34.1, 22.4, 5.9
As the frequencies always integer, therefore, by converting them to the nearest integers, we get
2, 10, 26, 34, 22, 6 are require frequencies.
56 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The following data shows that the number of seeds germinating out of 10 on damp filter for
80 set of seeds. Fit a Binomial distribution to the data.
x 0 1 2 3 4 5 6 7 8 9 10
f 6 20 28 12 8 6 0 0 0 0 0
Solution: Here the random variable X denotes the number of seeds germinating out of a set of 10
seeds.
The total number of trials is n = 10.
Evaluation of Mean
x f xf Probability P x f x N.P x
0 6 0 P x 0 0.086 6.88
1 20 20 P x 1 0.2390 19.12
2 28 56 P x 2 0.2990 23.92
3 12 36 P x 3 0.2216 17.73
4 8 32 P x 4 0.1078 8.62
5 6 30 P x 5 0.0360 2.88
6 0 0 P x 6 0.0083 0.67
7 0 0 P x 7 0.0013 0.11
8 0 0 P x 8 0.0001 0.01
9 0 0 P x 9 0.0000 0.00
10 0 0 P x 10 0.0000 0.00
N Σfi 80 Σfi xi 174
Σfi xi 174
Mean μ 2.175
Σfi 80
i.e., mean np 2.175
DSSM Compiled by: Magisetty Harikrushna 57
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
2.175
p 0.2175
10
q 1 p 1 0.2175 0.7825
n
Therefore, the Binomial distribution to be fitted as N q p
P x 0 n
C 0pxq n x
10 0 10 0
P x 0 C0 0.2175 0.7825
10
P x 0 0.7825 0.086
n x p
We have recurrence relation P x 1 .P x
x 1 q
10 0 0.2175
P 1 0.086 0.2390 ,
0 1 0.7825
10 1 0.2175
P 2 0.2390 0.2990 ,
1 1 0.7825
10 2 0.2175
P 3 0.2990 0.2216 ,
2 1 0.7825
10 3 0.2175
P 4 0.2216 0.1078 ,
3 1 0.7825
10 4 0.2175
P 5 0.1078 0.0360 ,
4 1 0.7825
10 5 0.2175
P 6 0.0360 0.0083 ,
5 1 0.7825
10 6 0.2175
P 7 0.0083 0.0013 ,
6 1 0.7825
10 7 0.2175
P 8 0.0013 0.0001 ,
7 1 0.7825
58 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
10 8 0.2175
P 9 0.0001 0.0000 ,
8 1 0.7825
10 9 0.2175
P 10 0.0000 0.0000 .
9 1 0.7825
The theoretical frequencies are 6.88, 19.12, 23.92, 17.73, 8.62, 2.88, 0.67, 0.11, 0.01, 0,0
As the frequencies always integer, therefore, by converting them to the nearest integers,
we get 7, 19, 24, 18, 9, 3, 1, 0, 0, 0 are require frequencies.
DSSM Compiled by: Magisetty Harikrushna 59
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
λxe λ
Poisson distribution: Define f(x) , x 0,1,2,3.... where λ is mean number of event
x!
occurrences within the particular time interval.
λxe λ
Clearly f(x) 0 and f(x)
x 0 x 0 x!
λ λx
e e λeλ 1
x 0 x!
Therefore f(x) satisfies the conditions of being a discrete type of pdf.
Definition: Any random variable X whose pdf is above f(x) is said to have Poisson distribution and
the pdf is called a Poisson pdf.
Here is called the parameter of Poisson distribution.
Note: Poisson distribution can be derived from Binomial distribution by applying the following
conditions
(i) p is very small (ii) n is large (iii) np λ (a finite value)
Derivation of Poisson distribution
The Poisson distribution can be derived from Binomial distribution with the conditions
(i) n is large
(ii) p is very small
λ
(iii) λ np p
n
x n− x
We have the Binomial distribution f ( x) = Cx p q , x = 0,1, 2...n,
n
n− x
x
f ( x) = C x
n
1 −
n n
n− x
n!
x
f ( x) = 1 −
x !( n − x )! n n
60 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
n
1−
n ( n − 1)( n − 2 ) ... ( n − ( x − 1) ) ( n − x )! x n
f ( x) =
x !( n − x )! nx x
1 −
n
1 2 x − 1 x
n
n 1 − 1 − ... 1 − 1 −
x
n n n n
f ( x) =
x
x !n x
1 −
n
1 2 x − 1 x
1 − 1 − ... 1 −
n
n
n n
f ( x) = 1 −
n
x
x !1 −
n
Taking as n →
1 2 x − 1 x
1 − 1 − ... 1 −
n
n
n n
lim f ( x) = lim 1 −
n
n → n → x
x !1 −
n
(1 − 0)(1 − 0) ...((1 − 0)) x − n
lim f ( x) = lim 1 +
x !(1 − 0) n
n→ x n→
x − a
x
f ( x) = e lim 1 + = ea
x! x →
x
x e−
Therefore, the Poisson distribution is f ( x) = , x = 0,1, 2...
x!
λxe λ
Mean of Poisson distribution: Mean μ xP(x) x
x 0 x 0 x!
λxe λ
x
x 0 x(x 1)!
DSSM Compiled by: Magisetty Harikrushna 61
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
λx e λ
x 1 (x 1)!
Put x 1 y x y 1,
limits x 1 y 0 and x y
λy 1e λ
μ
y 0 y!
λ λy
e λ
y 0 y!
λe λ e λ λ
Mean μ λ
2
Variance of the Poisson distribution: Variance σ x2 P(x) μ2
x 0
(x2 x x)P x μ2
x 0
x(x 1)P x (x)P x μ2
x 0 x 0
λx e λ
0 0 x(x 1) μ μ2 (x)P x μ
x 2 x(x 1)(x 2)! x 0
λx e λ
μ μ2
x 2 (x 2)!
λxe λ
σ2 μ μ2
x 2 (x 2)!
62 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Put x 2 y x y 2,
imits x 2 y 0 and x y
2 λy 2 e λ
σ μ μ2
y 0 y!
λ 2 λy
e λ μ μ2
y 0 y!
λ 2
e λ eλ μ μ2
λ2 λ λ2 λ
Therefore variance σ 2 λ
Standard deviation σ λ
μx e μ
Therefore, Poisson distribution function can be written as f(x) , x 0,1,2,3....
x!
Mode of the Poisson distribution
Mode is the value of x for which the probability p(x) is maximum.
Therefore p(x) p(x 1) and p(x) p(x 1)
λxe λ λx 1e λ
p(x) p(x 1)
x! (x 1)!
λx λx λ
x! (x 1)(x 1 1)!
1 λ
x! (x 1)(x)!
λ
1
(x 1)
(x 1) λ
DSSM Compiled by: Magisetty Harikrushna 63
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
x λ 1 (1)
And p(x) p(x 1)
λxe λ λx 1e λ
x! (x 1)!
1 λ1
x(x 1)! (x 1)!
1 1
x λ
x λ (2)
From (1) and (2) λ 1 x λ
Hence the mode of the Poisson distribution lies between λ 1 and λ
Note:
01. If λ is an integer, then λ 1 is also an integer. So, we have two maximum values and the
distribution is bimodal and two modes are λ and λ 1
02. If λ is not an integer, then mode of the distribution is integral part of λ
Recurrence relation of the Poisson distribution: P(x 1) xλ1 P(X)
P(x 1) λx 1e λ x! λx .λ.x! P(x 1) λ
P(X) (x 1)! λ e λ
x x
λ (x 1)x! P(X) x 1
64 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A hospital switch board received average of 4 emergency calls in a 10 min. interval, what
is the Probability that there are (i) At most 2 emergency calls in a 10 min. interval. (ii) Exactly 3
emergency calls in a 10 min. interval.
Solution: Given λ 4
(i) At most 2 emergency calls in a 10 min. interval
i.e., P( X 2) = P( X = 0) + P( X = 1) + P( X = 2)
0 e − 1e − 2e −
= + +
0! 1! 2!
4 41 4 2
e 1
1! 2!
P( X 2) = 13e−4
(ii) Exactly 3 emergency calls in a 10 min. interval
3e −
i.e., P( X = 3) =
3!
43
= e −4
3!
32 −4
P ( X = 3) = e
3
DSSM Compiled by: Magisetty Harikrushna 65
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: An insurance agent accepts polices of 5 men all of identical age and good in health. The
2
probability that a man of this age will be alive 30 years is . Find the probability that in 30 years (i)
3
All five men (ii) At least one man (iii) Almost three will be alive
2
Solution: Given n 5 and probability that a man will be alive 30 years is p
3
2 10
Therefore mean = = np = 5 =
3 3
μx e μ
The Poisson distribution is P(x) , x 0,1,2,3....
x!
(i) All five men will be alive
5 10
10 3
e
3
i.e., P(X 5)
5!
14.680
P(X 5)
120
P(X 5) 0.12
(ii) At least one man will be alive
i.e., P(X 1) 1 P X 1
P(X 1) 1 P X 0
0 10
10 3
e
3
P(X 1) 1
0!
10
P(X 1) 1 e 3
P(X 1) 1 0.0357
P(X 1) 0.9643
66 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
(iii) Almost three will be alive
i.e., P(X 3) 1 P x 3
P(X 3) 1 P x 4 P x 5
4 10 5 10
10 3 10 3
e e
3 3
1
4! 5!
P(X 3) 1 0.1835 0.12
P(X 3) 0.6965
Problem: Average number of accidents on any day on National highway is 1.8 () . Find the
probability that (i) At least once (ii) At most once
Problem: At a checkout counter customer arrive at an average of 1.5 per minute. Find the
probabilities that in any given minute of time (i) At most 4 will arrive (ii) Exactly 4 will arrive (iii)
At least 4 will arrive.
DSSM Compiled by: Magisetty Harikrushna 67
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The Poisson approximation to the binomial distribution
When the value of n in a Binomial distribution is large and the value of p is very small, the
binomial distribution can be approximated by a Poisson distribution.
If n 20 , p 0.05
Sometimes np 5 or n (1 − p ) 5 then the Poisson is a good approximation.
Problem: It has been found that 2% of the tools produced by certain machine are defective, what is
the Probability that in a shipment of 400 such tools (i) 3% or more (ii) 2% or less will be defective.
Solution: Given n = 400, p = 2% = 0.02
μ λ np
= = 400 0.02 = 8
Here n = 400 20 and p 0.05
Then Binomial distribution can be approximated by a Poisson distribution
x e−
Poisson distribution function f ( x) = , x = 0,1, 2,3....
x!
(i) 3% or more will be defective
i.e., P ( X 3%) = P(X 12) = 1 − P( X 12)
= 1 − {P(X = 0) + P(X = 1) + P(X = 2) + ... + P(X = 11)}
0 1 2 3 4 5 6 7 8 9 10 11
= 1 − e− + + + + + + + + + + +
0! 1! 2! 3! 4! 5! 6! 7! 8 9! 10! 11!
80 81 82 83 84 85 86 87 88 89 810 811
−8
= 1− e + + + + + + + + + + +
0! 1! 2! 3! 4! 5! 6! 7! 8 9! 10! 11!
= 1 − e−8 ( 2647.29 )
68 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
P(X 3%) 0.1119
(ii) 2% or less will be defective
i.e., P( X 2%) = P(X 8)
= {P(X = 0) + P(X = 1) + P(X = 2) + P(X = 3) + P(X = 4) + P(X = 5) + P(X = 6) + P(X = 7) + P(X = 8)}
0 e− 1e− 2 e− 3e − 4 e − 5e − 6 e − 7 e − 8e −
= + + + + + + + +
0! 1! 2! 3! 4! 5! 6! 7! 8
80 81 82 83 84 85 86 87 88
−8
=e + + + + + + + +
0! 1! 2! 3! 4! 5! 6! 7! 8
= e−8 (1766.3587 )
= 0.5925
DSSM Compiled by: Magisetty Harikrushna 69
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: In a given city, 6% of all drivers get at least one parking ticket per year. Use the Poisson
approximation to the binomial distribution to determine the probabilities that among 80 drivers: (i) 4
will get at least one parking ticket in any given year. (ii) At least 3 will get at least one parking ticket
in any given year. (iii) Anywhere from 3 to 6, inclusive, will get at least one parking ticket in any
given year.
Solution: Given n = 80 20, p = 6% = 0.06
= = np
= = 80 0.06 = 4.8
Here n = 80 20 and np 5
Then Binomial distribution can be approximated by a Poisson distribution
x e−
Poisson distribution function f ( x) = , x = 0,1, 2,3....
x!
(i) 4 will get at least one parking ticket in any given year.
( 4.8) e−4.8
4
i.e., P( X = 4) = = 0.182029
4!
(ii) At least 3 will get at least one parking ticket in any given year.
i.e., P( X 3) = 1 − P ( X 3)
1 P x 0 P x 1 P x 2
( 4.8)0 e−4.8 ( 4.8)1 e−4.8 ( 4.8)2 e−4.8
= 1− + +
0! 1! 2!
P ( X 3) = 1 − 0.142539
P( X 3) = 0.857461
(iii) Anywhere from 3 to 6, inclusive, will get at least one parking ticket in any given year
70 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
i.e., P(3 X 6) = P ( x = 3) + P ( x = 4 ) + P ( x = 5) + P ( x = 6 )
( 4.8) e−4.8 ( 4.8) e−4.8 ( 4.8) e−4.8 ( 4.8) e−4.8
3 4 5 6
P(3 X 6) = + + +
3! 4! 5! 6!
P (3 X 6) = 0.648265
Problem: If 0.8% of the fuses delivered to an arsenal are defective, use the Poisson approximation to
determine the probability that 4 fuses will be defective in a random sample of 400.
DSSM Compiled by: Magisetty Harikrushna 71
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A distributor of bean seeds determines from extensive tests that 5% of large batch of seeds
will not germinate. He sells the seeds in packets of 200 and guarantees 90% germination. Determine
the probability that a particular packet will violate the guarantee.
Solution: Let p be the probability of a seed not germinating, given p 5% 0.05
Here n 200 20
implies mean = = np = 200 0.05 = 10
A packet will violate guarantee if it contains more than 10% non-germination seeds
i.e., we have to evaluate P ( X 10%) = P ( X 20)
1 P(X 20)
20 λxe λ
1
x 0 x!
100 101 102 103 104 106 107 108 109 1010 1011 1012 1013 1014
+ + + + + + + + + + + + +
0! 1! 2! 3! 4! 6! 7! 8! 9! 10! 11! 12! 13! 14!
= 1− e
−10
1015 1016 1017 1018 1019 1020
+ + + + + +
15! 16! 17! 18! 19! 20!
1 0.9984
P(X 10%) 0.0016
72 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: If two cards are drawn from a pack of 52 cards which are diamonds, using Poisson
distribution find the probability of getting two diamonds at least 3 times in 51 consecutive trials of
two cards drawing each time.
Solution: Given n = 51
Let p be the probability of getting two diamonds from a pack of 52 cards.
13
C2 3
i.e., p
52 51
C2
3
Mean μ λ np 51 3
51
μx e μ
The Poisson distribution is P(x) , x 0,1,2,3....
x!
At least 3 times
i.e. P(X 3) 1 P(X 3)
1 P(X 0) P(X 1) P(X 2)
30 e 3
31 e 3
32 e 3 3 9
1 1 e 1 3
0! 1! 2! 2
P(X 3) 0.5768
DSSM Compiled by: Magisetty Harikrushna 73
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
2
Problem: If X is a random variable having Poisson distribution such that P(X 1) P(X 3) find (i)
3
Mean (ii) P(X 1) (iii) P(1 X 4)
2
Solution: Given X is a random variable having Poisson distribution such that P(X 1) P(X 3)
3
λ1e λ 2 λ3e λ
1! 3 3!
λ1 2 λ3
1 36
9λ λ3
λ3 9λ 0
λ λ2 9 0
λ 0 or λ 3
λ 3,0 are absurd values.
Mean μ λ 3
(ii) P(X 1) 1 P(X 1)
1 P(X 0)
λ0e λ
1
0!
3
P(X 1) 1 e
74 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
(iii) P(X 3) P(X 0) P(X 1) P(X 2) P(X 3)
λ0e λ λ1e λ λ2 e λ λ3e λ
0! 1! 2! 3!
3 31 3 2 3 3
e 1
1! 2! 3!
3
P X 3 13e
(iv) P(2 X 5) P(X 2) P(X 3) P(X 4) P(X 5)
λ2 e λ λ3e λ λ4e λ λ5e λ
2! 3! 4! 5!
3 32 33 3 4 35
e
2! 3! 4! 5!
72 3
P(2 X 5) e
5
DSSM Compiled by: Magisetty Harikrushna 75
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: If a Poisson distribution is such that 6P X 4 P X 2 2P X 0 , then find:
(i) Mean of x. (ii) P X 2 (iii) P X 2 .
Solution: Given 6P X 4 P X 2 2P X 0 ,
λ4e λ λ2 e λ λ0e λ
6 2
4! 2! 0!
λ4 λ2
6 2
24 2
λ4 λ2 4
4 2
λ4 2λ2 8 0
λ 1.849 2
(i) Mean μ λ 2
Problem: If X is a random variable having Poisson distribution such that P(X 1) P(X 2) find
(i) Mean (ii) P(X 1) (iii) P(X 3) (iv) P(2 X 5)
Problem: If a Poisson variate 2P(X 0) P(X 2) find the probability that (i) P(X 3) (ii) P(X 3)
(iii) P(2 X 5)
76 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Poisson variable has double mode at x 1 and x 2 , find the maximum probability and
also find P(X 2) .
Solution: Given Poisson distribution has double mode at x 1 and x 2
We know that Poisson is bimodal, and the two modes are at the points x λ 1 and x λ
Implies λ 2
μx e μ
The Poisson distribution is P(x) , x 0,1,2,3....
x!
The maximum probability is P(X 1) P(X 2)
21 e 2 2 2 e 2
1! 2!
2 2
2e 2e
2
4e
The maximum probability is 0.5413
And P(X 2) 1 P(X 2)
1 P(X 0) P(X 1)
20 e 2 21 e 2
1
0! 1!
2
1 3e
P(X 2) 0.5939
DSSM Compiled by: Magisetty Harikrushna 77
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Using recurrence relation for the Poisson distribution find the probabilities when
X 0,1, 2, 3, 4, 5. if the mean of the Poisson distribution is 3.
Solution: Given mean λ 3
λ0e λ 3
Now P(x 0) e
0!
λ
By recurrence relation we have P(x 1) P(X)
x 1
Put x 0
λ 3
P(0 1) P(0) 3e
x 1
Put x 1
λ 9 3
P(1 1) P(1) e
x 1 2
Put x 2
λ 9 3
P(2 1) P(2) e
x 1 2
Put x 3
λ 27 3
P(3 1) P(3) e
x 1 8
Put x 4
λ 81 3
P(4 1) P(4) e
x 1 40
78 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Fit a Poisson distribution to the following data
x 0 1 2 3 4 5 Total
f 142 156 69 27 5 1 400
Solution:
Evaluation of Mean
x f xf
0 142 0
1 156 156
2 69 138
3 27 81
4 5 20
5 1 5
Σfi 400 Σfi xi 400
Σfi xi 400
Mean μ λ 1
Σfi 400
The theoretical frequency for x successes is given by N.P(x) where X 0,1, 2, 3, 4, 5.
N.P(x) 400P(x 0), 400P(x 1), 400P(x 2), 400P(x 3), 400P(x 4), 400P(x 5)
10 e 1 11 e 1 12 e 1 13 e 1 14 e 1 15 e 1
400 , 400 , 400 , 400 , 400 , 400
0! 1! 2! 3! 4! 5!
147.15, 147.15, 73.58, 24.53, 6.13, 1.23
Therefore, the theoretical frequencies are 147, 147, 74, 25, 6, 1
Problem: Fit a Poisson distribution to the following data
x 0 1 2 3 4 5 6 7
f 305 365 210 80 28 9 2 1
DSSM Compiled by: Magisetty Harikrushna 79
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Normal distribution
The central limit theorem
The central limit theorem (CLT) states that the distribution of sample means approximates a normal
distribution as the sample size gets larger, irrespective of the population's distribution.
Or
If X is the mean of the random sample of size n taken from a population having the mean μ and the
x μ
finite variance σ 2 , then z is a random variable whose distribution function approaches that of
σ n
the standard normal distribution as n
Normal distribution, also known as the Gaussian distribution, is a probability distribution that
is symmetrical around its mean, ensuing in a bell-shaped curve. This distribution is widely used in
innumerable fields due to its mathematical properties and prevalence in natural phenomena.
Some real-life applications of the Normal Distribution
Finance and Economics
Stock prices: Changes in stock prices often follow a normal distribution pattern.
Returns on investments: The distribution of returns on investments is often modeled using a normal
distribution.
Quality Control
Manufacturing processes: Variations in product measurements, such as length, width, and weight,
often follow a normal distribution. Quality control charts are based on normal distribution
assumptions.
Biological Sciences
Physical characteristics: Traits like height, weight, and blood pressure in a population tend to follow
a normal distribution.
Biological variability: In studies of biological phenomena, the normal distribution is commonly used
to model genetic variations and other natural processes.
80 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Psychology
IQ scores: Intelligence quotient (IQ) scores are often assumed to follow a normal distribution in the
population.
Personality traits: Many personality traits, when measured, tend to follow a normal distribution.
Education
Standardized testing: Test scores on standardized exams, such as the SAT or GRE, are often
assumed to follow a normal distribution.
Grade distribution: In large populations, the distribution of grades in a course can approximate a
normal distribution.
Physics
Measurement errors: Errors in measurements and experimental data often follow a normal
distribution.
Quantum mechanics: Certain physical properties of particles, like their position and momentum, can
be modelled using normal distribution functions.
Social Sciences
Income distribution: In large populations, the distribution of income is often modeled using a log-
normal distribution, which is related to the normal distribution.
Survey responses: When collecting data through surveys, responses to questions can often be
analysed assuming a normal distribution.
Health Sciences
Blood pressure: Blood pressure measurements in a population tend to follow a normal distribution.
Body temperature: Normal body temperature variations in a healthy population can be modeled
using a normal distribution.
Environmental Studies
Weather patterns: Certain meteorological parameters, such as temperature and wind speed, can be
modelled using a normal distribution.
Pollution levels: Concentrations of pollutants in the air or water may follow a normal distribution.
DSSM Compiled by: Magisetty Harikrushna 81
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Engineering
Tolerance limits: Specifications for manufacturing parts often involve tolerance limits, and the
distribution of dimensions is assumed to be normal.
2
1 x a
1 2 b
Normal Distribution: Define f(x) e x
b 2π
Clearly f(x) 0, as b 0 and f(x)dx 1
Therefore f(x) satisfies the conditions of being a continuous pdf. This type of distribution is called
Normal distribution and the pdf f(x) is called the normal pdf.
Mean of Normal Distribution: Mean μ E(X) xf(x)dx
2
1 x a
1 2 b
μ E(X) x e dx
b 2π
2
1 x a
1 2 b
xe dx
b 2π
Put z x a x zb a dx bdz
b
if x z ,x z
z2
1 2
μ (zb a)e (bdz)
b 2π
z2
1 2
(zb a)e dz
2π
z2 z2
b 2 a 2
ze dz e dz
2π 2π
82 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
z2
b a 2
(0) 2 e dz
2π 2π 0
a a
f(x)dx 0, if f(x)isodd; 2 f(x)dx, if f(x) is even
a 0
z2
2a π π
a e 2 dz
2π 2 0 2
Mean μ a
Variance of the Normal Distribution: variance σ 2 x2 f(x)dx μ2
2
1 x a
1 2
x e 2 b
dx μ2
b 2π
x a
Put z b
x zb a dx bdz ; if x z ,x z
z2 z2
2 1 2 2 1 2
σ (zb a) e 2 (bdz) μ (zb a) e 2 dz μ2
b 2π 2π
z2 z2 z2
b2 2 2ab 2 1
z e 2 dz ze 2 dz a e 2 dz a2
2π 2π 2π
z2 z2
b2 2ab 1
2 z2 e 2 dz (0) a2 2 e 2 dz a2
2π 0 2π 2π 0
z2
2 b2 1 π
σ 2 z2 e 2 dz a2 2 a2
2π 0 2π 2
z2
b2
σ2 2 z2 e 2 dz a 2 a 2
2π 0
z2
2 2b2
σ z2 e 2 dz
2π 0
DSSM Compiled by: Magisetty Harikrushna 83
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
2b2 dt z2 dt
Now σ 2
(2t)e t
Put 2
t z2 2t 2zdz 2dt dz
2π 0 2t 2t
2b2 1
σ2 t 2 e t dt
π 0
2b2 3 1
t 2b 2 3 3 1
t 2 e dt Γ Γ 3
2
e t t 2 dt
π 0 π 2 0
2b2 1 1
σ2 Γ
π2 2
b2 1
π b2 Γ π
π 2
Variance σ 2 b2
Standard deviation σ b
2
1 x μ
1 2 σ
Now the Normal distribution pdf becomes f(x) e x
σ 2π
84 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Mode of the Normal Distribution
Mode is the value of x for which f(x) is maximum.
i.e., mode is the solution of f (x) 0 and f (x) 0
2
1 x μ
1 2 σ
We know that f(x) e
σ 2π
2
1 x μ
1 2 σ 1 x μ 1
f (x) e 2
σ 2π 2 σ σ
2
1 x μ
1 2 σ x μ
f (x) e (1)
σ 2π σ2
For point of maximum or minimum we take f (x) 0
2
1 x μ
1 2 σ x μ
e 0
σ 2π σ
x μ
0
σ
x μ
x μ
From (1) f (x) f(x)
σ2
x μ 1
f (x) f (x) f(x)
σ σ2
When x μ, f (x) 0
The point x μ is the point of maximum.
Therefore, the mode of the normal distribution is μ.
DSSM Compiled by: Magisetty Harikrushna 85
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Median of the Normal Distribution
M 1
If M is the median of the Normal distribution, we have f(x)dx f(x)dx
M 2
M 1
Now f(x)dx
2
2
M 1 x μ
1 2 σ 1
e dx
σ 2π 2
2 2
μ 1 x μ M 1 x μ
1 2 σ 1 2 σ 1
e dx e dx (1)
σ 2π σ 2π μ 2
2
μ 1 x μ
1 2 σ
Consider e dx
σ 2π
x μ
Put z then dx σdz
σ
If x z ; x μ z 0;
2
μ 1 x μ
1 2 σ
e dx
σ 2π
1 0 z2
e 2 σdz
σ 2π
1 0 z2 0 z2 π
e 2 dz e 2 dz
2π 2
1 π 1
2π 2 2
2
1 x μ
1 1 M 2 σ 1
(1) e dx
2 σ 2π μ 2
86 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
2
M 1 x μ
1 2 σ
e dx 0
σ 2π μ
2
M 1 x μ
2 σ
e dx 0
μ
b
Therefore μ M if f(x)dx 0 a b here f(x) 0
a
DSSM Compiled by: Magisetty Harikrushna 87
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Characteristics of the Normal Distribution:
01. The graph of the Normal distribution y f(x) in the xy – plane is known as the normal curve.
02. The curve is a bell-shaped curve and symmetrical with respect to mean
i.e., the line x μ and the two tails on the right and left sides of the mean extends to infinity. The
top of the bell is directly above the mean
03. Mean, median and mode of the distribution coincide at x μ as the distribution is symmetrical.
04. The probability that the normal variate X with mean μ and standard deviation lies between
x1 and x 2 is given by
2
x2 1 x μ
1 2 σ
P(x1 X x2 ) e dx
σ 2π x1
Standard Normal distribution: The Normal distribution with mean μ 0 and standard deviation σ 1
is known as the standard Normal distribution.
z1 z2
Note: The definite integral e 2 dz is known as the normal probability integral and gives the area
0
under standard normal curve between z 0 and z z1 is denoted by A(z1 )
88 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
How to find probability density of Normal Curve:
The probability that normal variate X with mean and standard deviation lies between two
specific values x1 and x 2 with x1 x2 can be obtained using area under the standard normal curve
as follows
x μ
01. Use transformation z & find z1 and z 2 corresponding to values x1 and x 2 respectively.
σ
02. To find P(x1 X x 2 ) P(z1 X z 2 )
Case (i) If both z1 and z 2 are positive (or both negative), then P(x1 X x2 ) A(z2 ) A(z1 )
= (Area under the normal curve from 0 to z2 ) − (Area under the normal curve from 0 to z1 )
DSSM Compiled by: Magisetty Harikrushna 89
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Case (ii) If z1 0 and z 2 0 (z1 0and z 2 0) then P(x1 X x 2 ) A(z 2 ) A(z1 )
= (Area under the normal curve from 0 to z2 ) + (Area under the normal curve from 0 to z1 )
03. To find P(x x1 ) P(z z1 )
P(z 0) P(z 0) 1
Case (i) If z1 0, then P(z z1 ) 0.5 A(z1 ) 2
90 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Case (ii) If z1 0, then P(z z1 ) 0.5 A(z1 )
04. To find P(x x1 ) P(z z1 )
Case (i) If z1 0, then P(z z1 ) 1 P(z z1 )
1 0.5 A(z1 )
P(z z1 ) 0.5 A(z1 )
DSSM Compiled by: Magisetty Harikrushna 91
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Case (ii) If z1 0, then P(z z1 ) 1 P(z z1 )
1 0.5 A(z1 )
P(z z1 ) 0.5 A(z1 )
Note: A( z) A(z).
92 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: If X is a normal variate with mean 30 & standard deviation 5, find the probability that (i)
26 X 40 (ii) X 45
Solution: Given that X is a normal variate with mean μ 30 and standard deviation σ 5.
x μ
We know that z
σ
x1 μ x2 μ
(i) Let z1 and z 2 where x1 26 and x 2 40
σ σ
26 30 40 30
z1 0.8 0 and z 2 2 0
5 5
Therefore P(x1 X x2 ) P(z1 z z 2 ) A(z 2 ) A(z1 )
P(26 X 40) P( 0.8 Z 2)
A( 0.8) A(2)
A(0.8) A(2)
0.4772 0.2881 (from normal tables)
0.7653
x1 μ 45 30
(ii) When x1 45 z1 3 0
σ 5
P(X x1 ) P(z z1 ) 0.5 A(z1 )
P(X 45) P(z 3) 0.5 A(3)
0.5 0.4987
P(X 45) 0.0013.
DSSM Compiled by: Magisetty Harikrushna 93
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: An electrical firm manufactures light bulbs that have a life, before burnout which is
normally distributed with mean equal to 800 hrs. and a standard deviation of 40 hrs. Find the
probability that a bulb burns between 778 hrs. and 834 hrs.
Solution: Given mean μ 800 and standard deviation σ 40
x μ
We have z
σ
x1 μ x2 μ
Now z1 and z 2 where x1 778 and x 2 834
σ σ
778 800 834 800
z1 0.55 0 and z 2 0.85 0
40 40
Therefore P(x1 X x2 ) P(z1 Z z2 ) A(z2 ) A(z1 )
i.e., P(778 X 834) P( 0.55 Z 0.85)
A(0.85) A( 0.55)
A(0.85) A(0.55)
0.3023 0.2088
P(778 X 834) 0.5111
94 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: In a sample of thousand cases the mean of a certain test is 14 and standard deviation is 2.5.
Assuming the distribution to be normal, how many students scored (i) Between 12 and 15
(ii) Above 18 (iii) Below 18.
Solution: Given that mean μ 14 and standard deviation σ 2.5
No. of students is n 1000
(i) Number of students scored between 12 and 15
i.e., P(12 X 15)
x μ
Here x1 12 & x 2 15. we know that z
σ
x1 μ 12 14 x 2 μ 15 14
z1 0.8 0 and z 2 0.4 0
σ 2.5 σ 2.5
Now P(x1 X x2 ) P(z1 X z 2 ) A(z1 ) A(z2 )
A( 0.8) A(0.4)
A(0.8) A(0.4)
0.2881 0.1554
0.4435.
For 1000 cases 1000 0.4435 443.5 444
(ii) The number of students scored above 18
i.e., P(X 18)
x1 μ 18 14
z1 z1 1.6 0
σ 2.5
P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A(1.6)
DSSM Compiled by: Magisetty Harikrushna 95
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
0.5 0.4452
0.0548
For 1000 cases 1000 0.0548 54.8 55
(iii) The number of students scored below 18
i.e., P(X 18)
x1 μ 18 14
z1 z1 1.6 0
σ 2.5
P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A(1.6)
0.5 0.4452
0.94452
For 1000 cases 1000 0.94452 945.2 945
96 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The mean and standard deviation of the marks obtained by 1000 students in an examination
are respectively 34.5 and 16.5. Assuming the normality of the distribution, find the approximate
number of students expected to obtain marks between 30 and 60.
Solution: Given mean μ 34.5 and standard deviation σ 16.5
Let x1 30 and x 2 60
x1 μ 30 34.5
We have z1 z1 0.27 0 and
σ 16.5
x2 μ 60 34.5
z2 z2 1.54 0
σ 16.5
Now P(x1 X x2 ) P(z1 X z 2 ) A(z1 ) A(z 2 )
i.e., P(30 X 60) P( 0.27 X 1.54)
A( 0.27) A(1.54)
A(0.27) A(1.54)
0.4382 0.1084
0.5916
For 1000 cases 1000 0.5916 591.6 592
DSSM Compiled by: Magisetty Harikrushna 97
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Suppose the weights of 800 male students are normally distributed with mean = 140
pounds and standard deviation 10 pounds. Find the number of students whose weights are (i)
Between 138 and 148 pounds (ii) More than 152 pounds.
Solution: Given mean = 140 and standard deviation = 10
(i) Let x1 138 and x2 148
x1 μ 140 138
We have z1 z1 0.2 0 and
σ 10
x2 μ 140 148
z2 z2 0.8 0
σ 10
P(x1 X x2 ) P(z1 X z 2 ) A(z1 ) A(z 2 )
i.e., P(138 X 148) P( 0.2 X 0.8)
A( 0.2) A(0.8)
A(0.2) A(0.8)
0.2881 0.0793
0.3674
Therefore, the number of students whose weights are between 138 and 148 pounds is
800 0.3674 293.9 294
(ii) Let x1 152
x1 μ 152 140
z1 z1 1.2 0
σ 10
P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A(1.2)
0.5 0.3849
0.1151
98 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Therefore, the number of students whose weights are more than 152 pounds is
800 0.1151 92.08 92
DSSM Compiled by: Magisetty Harikrushna 99
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A sample of 100 dry battery cells tested to find the length of life produced the following
results x 12, σ 3 hours. Assuming the data to be normally distributed, what percentage of battery
cells are expected to have life (i) More than 15 hours. (ii) Less than 6 hours. (iii) Between 10 and 14
hours?
Solution: Given that mean x 12 and standard deviation σ 3
(i) More than 15 hours
i.e., P(X 15)
x1 x 15 12
z1 z1 1 0
σ 3
P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A(1)
0.5 0.3413
0.1587
Percentage of battery cells are expected to have life more than 15 hours is 15.87%
(ii) Less than 6 hours.
i.e., P(X 6)
x1 x 6 12
z1 z1 2 0
σ 3
P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A( 2)
0.5 A(2)
0.5 0.4772
0.0228
100 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Percentage of battery cells are expected to have life less than 6 hours is 2.28%
(iii) Between 10 and 14 hours
i.e., P(10 X 14)
Here x1 10, x 2 14
x x
we know that z
σ
x1 x 10 12 x 2 μ 14 12
z1 0.67 0 and z 2 0.67 0
σ 3 σ 3
Now P(x1 X x2 ) P(z1 X z 2 ) A(z1 ) A(z2 )
A( 0.67) A(0.67)
A(0.67) A(0.67)
2A(0.67)
2 0.2485
0.4970
Percentage of battery cells are expected to have life between 10 and 14 hours is 49.70%
DSSM Compiled by: Magisetty Harikrushna 101
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: The life of army shoes is normally distributed with mean 8 months and standard deviation
2 months. If 5000 pairs are insured, how many pairs would be expected to need replacement after 12
x x
months. Given that P Z 2 0.0228 and z
σ
Solution: Given total number pairs is n 5000
Mean x 8 months and standard deviation σ 2 months. P x 12 P Z 2 0.5 A Z
Total months x 12 P x 12 0.5 A 2
x x 12 8 P x 12 0.5 0.4772
When x 12 , z z 2 0
σ 2
P x 12 0.9772
Now P x 12 P Z 2 0.0228
Total pairs need to replace is
Number of pairs whose life is more than 12 months is 5000 0.0228 114
Therefore, the pairs of shoes need to replace after 12 months is 5000 114 4886
Problem: The mean yield per plot of a crop is 17 kg. and standard deviation is 3 kg. If distribution of
yield per plot is normal, find the percentage of plot giving yield (i) Between 15.5 kg. and 20 kg. (ii)
More than 20 kg.
Ans (i) 53.28% (ii) 15.87%
Problem: If the masses of 300 students are normally distributed with mean 68 kg. and standard
deviation 3kg. How many students have masses,
(i) Less than or equal to 72 kg. ( 28 )
(ii) Less than or equal to 64 kg. ( 28 )
(iii) Between 65 and 71 kg. ( 205 )
102 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The marks obtained in mathematics by 1000 students are normally distributed with mean
78% and standard deviation 11%. Determine: (i) How many students got marks above 90%.
(ii) What was the highest mark obtained by the lowest 10% of the students? (iii) With in what
limit did the middle of 90% of the students lie.
Solution: Given that mean μ 78% 0.78 and standard deviation σ 11% 0.11
(i) Marks above 90%.
i.e., P(X 0.9)
x1 μ 0.9 0.78
z1 z1 1.09 0
σ 0.11
P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A(1.09)
0.5 0.3621
0.1379
Hence the number of students with marks more than 90% is 0.1379 1000 137.9 138
(ii) Highest mark obtained by the lowest 10% of the students
The 0.1 area to the left of z corresponds to the lowest 10% of the students
From diagram 0.4 0.5 0.1
i.e., A z1 0.4
Implies z1 1.28
x1 μ x 0.78
z1 1.28 x 0.6392
σ 0.11
Hence the height marks obtained by the lowest 10% of students 0.6392 1000 64%
(iii) Middle 90% corresponds to 0.9 area, leaving 0.05 area on both sides, then form diagram
DSSM Compiled by: Magisetty Harikrushna 103
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Now the area from 0 to z 1 is A z1 0.45
z1 1.64 and z2 1.64
x μ
we know that z
σ
x1 μ x1 0.78
z1 1.64 x1 0.5996 or 59.96% and
σ 0.11
x2 μ x 2 0.78
z2 1.64 x 2 0.9604 or 96.04%
σ 0.11
104 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: If X is a normal variate, find the area to the (i) Left of z 1.78 (ii) Right of z 1.45
(iii) Corresponding to 0.8 Z 1.58 (iv) to the left of z 2.52 and to the right of z 1.83
Solution: (i) To the left of z 1.78 0
P(Z z1 ) 0.5 A(z1 )
0.5 A( 1.78)
0.5 A(1.78)
0.5 0.4625
P(Z z1 ) 0.0375
(ii) To the right of z 1.45 0
P(Z z1 ) 0.5 A(z1 )
0.5 A( 1.45)
0.5 A(1.45)
0.5 0.4265
P(Z z1 ) 0.9265
(iii) P( 0.8 Z 1.53)
A( 0.8) A(1.53)
A(0.8) A(1.53)
0.2881 0.43701
P( 0.8 Z 1.53) 0.7251
DSSM Compiled by: Magisetty Harikrushna 105
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
(iv) To the left of z 2.52
i.e., P(Z z1 ) 0.5 A(z1 )
0.5 A( 2.52) 0.5 0.4971
0.0059
To the right of z 1.83
i.e., P(Z z1 ) 0.5 A(z1 )
0.5 A(1.83) 0.5 0.4664
P(Z z1 ) 0.0336
To the left of z 2.52 and to the right of z 1.83 is P(Z z1 ) P(Z z1 )
i.e., 0.0059 0.336 0.0395
106 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Suppose 10% of the students Probability for a Normal distribution is below 35 and 5%
above 90. What are values of μ and σ .
Solution: Given that P(X 35) 10% 0.1 and P(X 90) 5% 0.05
P(X 35) P(X x1 ) P(Z z1 ) 0.1
From graph P(z1 Z 0) 0.5 0.1 0.4
i.e., A z1 0.4
z1 1.29 (From tables)
P(X 90) P(X x 2 ) P(Z z 2 ) 0.05
From graph P(0 Z z 2 ) 0.5 0.05 0.45
i.e., A z 2 0.45
z2 1.65 (From tables)
x1 μ
We know that z1
σ
35 μ
1.29
σ
1.29σ 35 μ
1.29σ μ 35 (1) and
x2 μ
z2
σ
90 μ
1.65
σ
1.65σ 90 μ
1.65σ μ 90 (2)
DSSM Compiled by: Magisetty Harikrushna 107
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
By solving (1) and (2)
we get Standard Deviation σ 18.70 and Mean μ 59.1
108 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: In a normal distribution 7% of items under 35 and 89% of items under 63. What are the
mean and standard deviation of the distribution?
Solution: Given P(X 35) 7% 0.07& P(X 63) 89% 0.89
P(X 35) 7% 0.07 and
P(X 63) 89% 0.89
P(X 63) 1 P(X 63)
1 0.89 0.11
i.e. P(X 35) 7% 0.07 and P(X 63) 0.11
Now P(X 35) P(X x1 ) P(Z z1 ) 0.07
From graph P(z1 Z 0) 0.5 0.07 0.43
i.e., A z1 0.43
z1 1.48 (From tables)
P(X 63) P(X x 2 ) P(Z z 2 ) 0.11
From graph P(0 Z z 2 ) 0.5 0.11 0.39
i.e., A z 2 0.39
z2 1.23 (From tables)
x1 μ
We know that z1
σ
35 μ
1.48
σ
1.48σ 35 μ
1.48σ μ 35 (1) and
DSSM Compiled by: Magisetty Harikrushna 109
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
x2 μ
z2
σ
63 μ
1.23
σ
1.23σ 63 μ
1.23σ μ 63 (2)
By solving (1) and (2) we get σ 10.332 and μ 50.3
110 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The marks obtained in statistics in a certain examination found to be normally distributed,
if 15% of the students greater than 60 marks, 40% of the students less than 30 marks then find μ & σ
Solution: Given that P(X 30) 40% 0.4 and P(X 60) 15% 0.15
P(X 30) P(X x1 ) P(Z z1 ) 0.4
From graph P(z1 Z 0) 0.5 0.4 0.1
i.e., A z1 0.1
z1 0.26 (From tables)
P(X 60) P(X x 2 ) P(Z z 2 ) 0.15
From graph P(0 Z z 2 ) 0.5 0.15 0.35
i.e., A z 2 0.35
z2 1.04 (From tables)
x1 μ
We know that z1
σ
30 μ
0.26
σ
0.26σ 30 μ (1) and
x2 μ
z2
σ
60 μ
1.04
σ
1.04σ 60 μ (2)
By solving (1) and (2)
we get = 23.076 and = 36
DSSM Compiled by: Magisetty Harikrushna 111
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: In a normal distribution 31% of the items are under 45 and 8% are over 64. Find the mean
and variance of the distribution.
112 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Binomial, Poisson approximation to the Normal distribution
To calculate the probabilities with large values of n, you had to use the binomial formula,
which could be very complicated. Using the normal approximation to the binomial distribution
simplified the process.
If we have Binomial distribution with parameter n, p , if np 5, nq 5 then the Binomial
distribution can be approximated to Normal distribution with mean = np and variance = npq .
2
If we have Poisson distribution with parameter , if 20, then the Poisson distribution can be
approximated to Normal distribution with mean = and variance = .
2
Note: The Poisson distribution is discrete distribution, where as Normal distribution is continuous
distribution. While changing discrete distribution to continues distribution requires continuity
correction factor ( −0.5 or + 0.5 ) .
Discrete Distribution Normal Distribution Discrete Distribution Normal Distribution
P ( X 6) P ( X 6.5) P ( 6 X 9) P ( 5.5 X 9.5)
P ( X 6) P ( X 5.5) P ( 6 X 9) P ( 6.5 X 9.5)
P ( X 6) P ( X 5.5) P ( 6 X 9) P ( 5.5 X 8.5)
P ( X 6) P ( X 6.5) P ( 6 X 9) P ( 6.5 X 8.5)
DSSM Compiled by: Magisetty Harikrushna 113
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: 12 Coins are tossed together. Find the probability of getting to 8 heads in single toss, using
Normal approximation.
Solution: Let X be the variable of getting head of tossing a coin.
1 1
Here n = 12 , p is the probability of getting head, p = q=
2 2
1 1
Mean = np = 12 = 6 5 and nq = 12 = 6 5
2 2
1 1
Variance = npq = 12 = 3
2
2 2
Standard deviation = 3 = 1.732
1
Now X B 12, is approximated to X N ( = 6, 2 = 3)
2
The probability of getting to 5 heads in single toss is P (1 X 8)
By applying continuity correction factor
P (1 X 8) = P ( 0.5 X 8.5)
Here x1 0.5, x 2 8.5
x x
we know that z
σ
x1 μ 0.5 6 x 2 μ 8.5 6
z1 3.18 0 and z 2 1.44 0
σ 1.732 σ 1.732
Now P(x1 X x2 ) P(z1 Z z2 ) A(z1 ) A(z2 )
A( 3.18) A(1.44) A(3.18) A(1.44)
0.4993 0.4251
P 1 X 8 0.9244
114 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: In a city, 46 percent of the population favour the incumbent, Morgan, for mayor. A simple
random sample of 500 is taken. Using the continuity correction factor, find the probability that at
least 250 favour Morgan for mayor.
Solution: Let X be the variable that population favour the incumbent.
Here n = 500 , p is the probability of the population favour the incumbent,
p = 46% = 0.46 q = 0.54
Mean = np = 500 ( 0.46 ) = 230 5 and nq = 500 ( 0.54 ) = 270 5
Variance = npq = 500 ( 0.46 )( 0.54 ) = 124.2
2
Standard deviation = 124.2 = 11.1445
Now X B ( 500,0.46 ) is approximated to X N ( = 230, 2 = 124.2 )
The probability that at least 250 favour Morgan for mayor is P ( X 250 )
By applying continuity correction factor
P ( X 250 ) = P ( X 249.5)
Here x1 249.5
x x
we know that z
σ
x μ 249.5 230
z1 1.75 0
σ 11.1445
Now P(X x1 ) P(Z z1 ) 0.5 A(z1 )
0.5 A(1.75)
0.5 0.4599
P X 250 0.0401
DSSM Compiled by: Magisetty Harikrushna 115
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Find the probability by guess work a student can correctly answer 25 to 30 questions out of
80 questions. In each question with four choices, only one is correct and the student has no
knowledge on the subject.
Solution: Let X be the variable that correct answer.
1 3
Here n = 80 , p is the probability of choice, p = q=
4 4
1 3
Mean = np = 80 = 20 5 and nq = 80 = 60 5
4 4
Variance = npq = 80 ( 0.25)( 0.75) = 15
2
Standard deviation = 15 = 3.872
1
Now X B 80, is approximated to X N ( = 20, 2 = 15)
4
The probability by guess work a student can correctly answer 25 to 30 questions is P ( 25 X 30 )
By applying continuity correction factor
P ( 25 X 30 ) = P ( 24.5 X 30.5)
Here x1 24.5, x 2 30.5
x x
we know that z
σ
x1 μ 24.5 20 x 2 μ 30.5 20
z1 1.16 0 and z 2 2.71 0
σ 3.872 σ 3.872
Now P(x1 X x2 ) P(z1 Z z2 ) A(z2 ) A(z1 )
A(2.71) A(1.16)
0.4966 0.3770
P 25 X 30 0.1196
116 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The average number of calls received by a hotel receptionist is 6 per hour. Find the
probability of receiving (i) Less than 17 calls in 4 hours. (ii) More than 31 calls in 6 hours.
Solution: Let X be the variable that number of calls received.
Given the average number of calls received by a hotel receptionist is 6 per hour.
Mean = 6
i.e., X Pos ( 6 )
(i) Less than 17 calls in 4 hours.
Now the mean value for 4 hours is = 4 = 24 20
Now X Pos ( 24 ) is approximated to X N ( = 24, 2 = 24)
Implies standard deviation = 24 = 4.899
Now we have to find P ( X 17 )
By applying continuity correction factor
P ( X 17 ) = P ( X 16.5)
x−
We have z =
16.5 − 24
z= = −1.53 0
4.899
Now P ( X x1 ) = P ( Z z1 ) = 0.5 − A ( Z )
P ( Z −1.53) = 0.5 − A ( −1.53)
P ( Z −1.53) = 0.5 − A (1.53)
P ( Z −1.53) = 0.5 − 0.4370
DSSM Compiled by: Magisetty Harikrushna 117
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
P ( X 16.5) = P ( Z −1.53) = 0.0630
P ( X 17 ) = P ( X 16.5) = 0.0630
(ii) More than 31 calls in 6 hours.
Now the mean value for 6 hours is = 6 = 36 20
Now X Pos ( 36 ) is approximated to X N ( = 36, 2 = 36 )
Implies standard deviation = 36 = 6
Now we have to find P ( X 31)
By applying continuity correction factor
P ( X 31) = P ( X 31.5)
x−
We have z =
31.5 − 36
z= = −0.75 0
6
Now P ( X x1 ) = P ( Z z1 ) = 0.5 + A ( Z )
P ( X 31.5) = P ( Z −0.75) = 0.5 + A ( −0.75)
P ( X 31.5) = P ( Z −0.75) = 0.5 + A ( 0.75)
P ( X 31.5) = P ( Z −0.75) = 0.5 + 0.2734
P ( X 31.5) = P ( Z −0.75) = 0.7734
P ( X 31) = 0.7734
118 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Per hour there are an average of 25 buses that passes the central station. What is the
probability that exactly 25 will pass bus stop central station between 10 am and 11 am.
Solution: Let X be the variable that number of buses that passes the central station.
Given the average number of buses that passes the central station is 25.
Mean = = 25
Now X Pos ( 25) is approximated to X N ( = 25, 2 = 25)
Implies standard deviation = 25 = 5
Probability that exactly 25 is P ( X = 25)
P ( X = 25) = P ( 24 X 26 )
By applying continuity correction factor
P ( X = 25) = P ( 24.5 X 25.5)
Here x1 24.5, x 2 25.5
x x
we know that z
σ
x1 μ 24.5 25 x 2 μ 25.5 25
z1 0.1 0 and z 2 0.1 0
σ 5 σ 5
Now P(x1 X x2 ) P(z1 Z z2 ) A(z1 ) A(z2 )
A( 0.1) A(0.1)
A(0.1) A(0.1)
2A(0.1)
2 0.0398
P X 25 0.0796
DSSM Compiled by: Magisetty Harikrushna 119
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: The number of visitors to an Internet website in a day follows a Poisson distribution. On
average there are 30 visitors in a given day. Use the normal approximation to the Poisson distribution
to find the probability of the website receiving between 25 and 32 visitors inclusive in a given day.
Solution: Let X be the variable that number of visitors.
Given the average there are 30 visitors in a given day.
Mean = = 30
Now X Pos ( 30 ) is approximated to X N ( = 30, 2 = 30 )
Implies standard deviation = 30 = 5.477
The probability of the website receiving between 25 and 32 visitors inclusive in a given day.
i.e., P ( 25 X 32 )
By applying continuity correction factor
P ( 25 X 32 ) = P ( 24.5 X 32.5)
Here x1 24.5, x 2 32.5
x x
we know that z
σ
x1 μ 24.5 30 x 2 μ 32.5 30
z1 1.01 0 and z 2 0.46 0
σ 5.477 σ 5.477
Now P(x1 X x2 ) P(z1 Z z2 ) A(z1 ) A(z2 )
A( 1.01) A(0.46)
A(1.01) A(0.46)
0.3438 0.1772
P 25 X 32 0.521
120 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Uniform distribution
A continuous random variable X is said to have a Uniform distribution over the interval a,b
k a x b
, if its probability density function is given by f ( x ) = .
0 other wise
b
We have f ( x ) dx = 1
a
b
k dx = 1
a
b
k 1dx 1
a
b
k x a
1
k b a 1
1
k
b a
1
a xb
Then the probability density function is f ( x ) = b − a
0 other wise
x
The cumulative distribution function is F ( x ) = P ( X x ) = f ( x ) dx
−
x
When x a then F ( x ) = P ( X x ) = f ( x ) dx = 0
−
b
x
When a x b then F ( x ) = P ( X x ) = f ( x ) dx
−
a x
F ( x) = f ( x ) dx + f ( x ) dx
− a
DSSM Compiled by: Magisetty Harikrushna 121
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
x
1
F ( x) = 0 + dx
a
b − a
x
1
F ( x) =
b − a a
1 dx
1
F ( x) = ( x )a
x
b−a
x−a
F ( x) =
b−a
x
When x b then F ( x ) = P ( X x ) = f ( x ) dx
−
a b x
F ( x) = f ( x ) dx + f ( x ) dx + f ( x ) dx
− a b
b
1
F ( x) = 0 + dx + 0
a
b−a
b
1
F ( x) =
b − a a
1 dx
1
F ( x) = ( x )a
b
b−a
b−a
F ( x) = =1
b−a
0 if x a
x−a
Therefore, cumulative distribution function is F ( x ) = P ( X x ) = if a x b
b − a
1 if x b
122 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
b
Mean: Mean = E ( x ) = xf ( x ) dx
a
b
1
=x dx
a
b−a
b
1
b − a a
= x dx
b
1 x2
=
b − a 2 a
1 b2 − a 2
=
b−a 2
1 ( b − a )( b + a )
=
b−a 2
= E ( x) =
(a + b)
2
b
Variance: Variance 2 = x 2 f ( x ) dx − 2
a
b
1
2 = x2 dx − 2
a
b−a
b
1
2 =
b−a a
x 2 dx − 2
b
1 x3
=
2
−
2
b − a 3 a
1 b3 − a 3 a + b
2
2 = −
b−a 3 2
2 =
2
(
1 ( b − a ) b + ab + a
2
) − a + b 2
b−a 3 2
DSSM Compiled by: Magisetty Harikrushna 123
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
b 2 + ab + a 2 a 2 + 2ab + b 2
2 = −
3 4
a 2 − 2ab + b 2
2 =
12
Variance 2 =
( b − a )2
12
124 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
x 2
Problem: A random variable X has the pdf f ( x ) = 4 find (i) P ( X 1)
0 other wise
(ii) P ( X 1) (iii) P ( 2 X + 3 5)
1
−2 x 2
Solution: Given X has a uniform distribution f ( x ) = 4
0 other wise
0 if x a
x−a
The cumulative distribution function is F ( x ) = P ( X x ) = if a x b
b − a
1 if x b
0 if x −2
x+2
i.e., F ( x ) = if − 2 x 2
4
1 if x 2
(i) P ( X 1)
1
P ( X 1) = f ( x ) dx
−
−2 1
P ( X 1) = f ( x ) dx + f ( x ) dx
− −2
−2 1
1
P ( X 1) = 0 dx + 4 dx
− −2
1
1
P ( X 1) =
4 −2
1 dx
1
P ( X 1) = ( x )1−2
4
1
P ( X 1) = (1 + 2 )
4
DSSM Compiled by: Magisetty Harikrushna 125
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
3
P ( X 1) =
4
(ii) P ( X 1)
P ( X 1) = 1 − P ( X 1)
P ( X 1) = 1 − P ( −1 X 1)
1
P ( X 1) = 1 − f ( x ) dx
−1
1
P ( X 1) = 1 −
1
dx
−1
4
1
P ( X 1) = 1 − 1 dx
1
4 −1
P ( X 1) = 1 −
1 1
( x )−1
4
P ( X 1) = 1 −
1
(1 + 1)
4
P ( X 1) = 1 −
1 1
=
2 2
(iii) P ( 2 X + 3 5)
P ( 2 X + 3 5) = P ( 2 X 2 )
P ( 2 X + 3 5) = P ( X 1)
P ( 2 X + 3 5) = f ( x ) dx
1
2
P ( 2 X + 3 5) = f ( x ) dx + f ( x ) dx
1 2
2
1
P ( 2 X + 3 5) = dx + 0 dx
1
4 2
126 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
2
1
P ( 2 X + 3 5) =
4 1
1 dx
1
P ( 2 X + 3 5) = ( x )1
2
4
1
P ( 2 X + 3 5) = ( 2 − 1)
4
1
P ( 2 X + 3 5) =
4
DSSM Compiled by: Magisetty Harikrushna 127
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
Problem: If X is a uniform distribution in −2, 2 , find (i) P ( X 0 ) (ii) P X − 1
2
Solution: Given X is a uniform distribution in −2, 2
1
−2 x 2
The Probability distribution function is f ( x ) = 4
0 other wise
0 if x a
x−a
The cumulative distribution function is F ( x ) = P ( X x ) = if a x b
b − a
1 if x b
0 if x −2
x+2
i.e., F ( x ) = if − 2 x 2
4
1 if x 2
(i) P ( X 0 )
0
P ( X 0) = f ( x ) dx
−
−2 0
P ( X 0) = f ( x ) dx + f ( x ) dx
− −2
−2 0
1
P ( X 0) = 0 dx + 4 dx
− −2
1
(ii) P X − 1
2
1 1
P X −1 = 1 − P X −1
2 2
1 1 1
P X −1 = 1 − P − X −1
2 2 2
1 1 1
P X − 1 = 1 − P 1 − X 1 +
2 2 2
128 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 1 3
P X −1 = 1 − P X
2 2 2
3
1 2
P X − 1 = 1 − f ( x ) dx
2 1
2
1 2
1
P X − 1 = 1 − dx
2 1 4
2
1 2
1
P X − 1 = 1 − 1 dx
2 41
2
3
1 1
P X − 1 = 1 − ( x ) 12
2 4 2
1 13 1
P X −1 = 1 − −
2 4 2 2
1 1
P X − 1 = 1 − (1)
2 4
1 3
P X −1 =
2 4
DSSM Compiled by: Magisetty Harikrushna 129
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: The cumulative distribution function of a random variable X is given by
0 if x −
x +
F ( x) = if − x Find (i) The probability density function
2
1 if x
−
(ii) P X
2 2
0 if x −
x +
Solution: Given the cumulative distribution function F ( x ) = if − x
2
1 if x
d
(i) We know that f ( x ) = F ( x)
dx
0 if x −
1
f ( x) = if − x
2
0 if x
1
if − x
i.e., f ( x ) = 2
0 else where
−
(ii) P X
2 2
By cumulative distribution function P ( a X b ) = F ( b ) − F ( a )
− −
P X = F −F
2 2 2 2
+ − +
−
P X = 2 − 2
2 2 2 2
3
− 2
P X = − 2
2 2 2 2
130 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
− 3 1
P X = −
2 2 4 4
− 1
P X =
2 2 2
Problem: If X is uniform variable in −1, 1 , find the probability density function, mean and variance.
DSSM Compiled by: Magisetty Harikrushna 131
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A random variable X has a uniform pdf with T = 10, find the probability
(i) 1 X 3 (ii) 2 X 5.5 (iii) X 2.9 (iv) X 6.8 (v) 6.2 X 8.6
Solution: Given X has a uniform distribution
1
a xb
The probability density function is f ( x ) = b − a
0 other wise
Here given T = b − a = 10
(i) P (1 X 3)
3
P (1 X 3) = f ( x ) dx
1
3
1
P (1 X 3) = dx
1
10
3
1
P (1 X 3) = 1 dx
10 1
1
P (1 X 3) = ( x )13
10
1 1
P (1 X 3) = ( 3 − 1) =
10 5
P (1 X 3) = 0.2
(ii) P ( 2 X 5.5)
5.5
P ( 2 X 5.5 ) = f ( x ) dx
2
5.5
1
P ( 2 X 5.5 ) = 10 dx
2
132 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
5.5
1
P (1 X 3) = 1 dx
10 2
1
P (1 X 3) = ( x )5.5
2
10
1 3.5
P (1 X 3) = ( 5.5 − 2 ) =
10 10
P (1 X 3) = 0.35
(iii) P ( X 2.9 )
10
P ( X 2.9 ) = f ( x ) dx
2.9
10
1
P ( X 2.9 ) = 1 dx
10 2.9
1
P ( X 2.9 ) = ( x )102.9
10
1
P ( X 2.9 ) = (10 − 2.9 )
10
7.1
P ( X 2.9 ) =
10
P ( X 2.9 ) = 0.71
(iv) P ( X 6.8)
6.8
P ( X 6.8 ) = f ( x ) dx
0
6.8
1
P ( X 6.8 ) = 10 dx
0
1
P ( X 6.8 ) = ( x )6.8
0
10
DSSM Compiled by: Magisetty Harikrushna 133
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
P ( X 6.8 ) = ( 6.8 − 0 )
10
P ( X 6.8) = 0.68
(v) 6.2 X 8.6
8.6
P ( 6.2 X 8.6 ) = f ( x ) dx
6.2
8.6
1
P ( 6.2 X 8.6 ) = 10 dx
6.2
8.6
1
P ( 6.2 X 8.6 ) = 1 dx
10 6.2
1
P ( 6.2 X 8.6 ) = ( x )8.6
6.2
10
1 2.4
P ( 6.2 X 8.6 ) = (8.6 − 6.2 ) =
10 10
P ( 6.2 X 8.6 ) = 0.24
134 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Suppose X is a uniform random variable in the interval ( −3, 3) and Y = X 2 . Obtain the
cumulative distribution function of the random variable Y. Also calculate the mean and variance of
the random variable X.
Solution: Given X is a uniform random variable in the interval ( −3, 3)
1
a xb
The Probability distribution function is f ( x ) = b − a
0 other wise
1
−3 x 3
i.e., f ( x ) = 6
0 other wise
0 if x a
x−a
The cumulative distribution function is F ( x ) = P ( X x ) = if a x b
b − a
1 if x b
0 if x −3
x+3
i.e., F ( x ) = if − 3 x 3
6
1 if x 3
Given Y = X 2
X = Y
−3 X 3 0 X 2 9
i.e., 0 Y 9
The cumulative distribution function of Y is F ( y ) = P (Y y )
F ( y) = P ( X 2 y)
(
F ( y) = P − y X y )
DSSM Compiled by: Magisetty Harikrushna 135
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
y
F ( y) = f ( x ) dx
− y
y
1
F ( y) = dx
− y
6
y
1
F ( y ) = 1 dx
6− y
1
F ( y) = ( x )− y
y
F ( y) =
1
6
( y+ y )
y
F ( y) =
3
0 if y 0
y
The cumulative distribution function of Y is F ( y ) = P (Y y ) = if 0 y 9
3
1 if y 9
Mean of X is = E ( X ) = x f ( x ) dx
−
3
1
= E ( X ) = x dx
−3
6
3
1
= E ( X ) = x dx
6 −3
a
2 f ( x ) dx if f ( x ) is even function
f ( x ) dx 0
a
1
= E(X ) = (0)
6
−a
0 if f ( x ) is odd function
Mean of X is = 0
Or
136 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
a + b −3 + 3
Mean = = =0
2 2
Variance of X is = x f ( x ) dx −
2 2 2
−
3
1
= x2 dx − ( 0 )
2 2
−3
6
3
1
= x 2 dx
2
6 −3
a
2 f ( x ) dx if f ( x ) is even function
f ( x ) dx 0
3 a
2 2
2 =
6 0
x dx
−a 0 if f ( x ) is odd function
3
1 x3
=
2
3 3 0
1
2 = ( 27 − 0 )
9
Variance of X is 2 = 3
Or
(b − a ) ( 3 + 3)
2 2
36
Variance = 2
= = =3
12 12 12
DSSM Compiled by: Magisetty Harikrushna 137
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: If X is a uniform random variable in the interval ( 0, 1) and Y = −2 log X , Obtain the
cumulative distribution function of the random variable Y. Also calculate the mean and variance of
the random variable X.
Solution: Given X is a uniform random variable in the interval ( 0, 1)
1
a xb
The Probability distribution function is f ( x ) = b − a
0 other wise
1 0 x 1
i.e., f ( x ) =
0 other wise
0 if x a
x−a
The cumulative distribution function is F ( x ) = P ( X x ) = if a x b
b − a
1 if x b
0 if x 0
i.e., F ( x ) = x if 0 x 1
1 if x 1
Given Y = −2 log X , 0 X 1
0 X 1 − log X 0
0 −2 log X
i.e., 0 Y
The cumulative distribution function of Y is F ( y ) = P (Y y )
F ( y ) = P ( −2log X y )
y
F ( y ) = P log X
−2
−
y
F ( y) = P X e 2
138 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
−
y
F ( y) = 1− P X e
2
y
−
e 2
F ( y) = 1− f ( x ) dx
−
y
−
e 2
F ( y) = 1− f ( x ) dx
0
y
−
e 2
F ( y) = 1− 1 dx
0
y
−
F ( y ) = 1 − ( x )0
e 2
− 2y
F ( y ) = 1 − e − e0
y
−
F ( y) = 2 − e 2
0 if y 0
The cumulative distribution function is F ( y ) = −
y
2 − e 2 if 0 y
Mean of X is = E ( X ) = x f ( x ) dx
−
1
= E ( X ) = x .1 dx
0
1
= E ( X ) = x dx
0
1
x2
= E(X ) =
2 0
1
= E ( X ) = − 0
2
DSSM Compiled by: Magisetty Harikrushna 139
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
Mean of X is =
2
Variance of X is = x f ( x ) dx −
2 2 2
−
2
1
1
= x .1 dx −
2 2
0 2
1
x3 1
= −
2
3 0 4
1 1
2 = − 0 −
3 4
1 1
2 = −
3 4
1
Variance of X is 2 =
12
140 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Exponential Distribution
A continuous random variable X is said to have exponential distribution with parameter , if
e − x x0
X has the probability density function f ( x ) = , it is denoted by X Exp ( ) .
0 else where
Mean:
The mean of the exponential distribution is = E ( x ) = x f ( x ) dx
−
= E ( x ) = x e− x dx
0
= E ( x ) = x e− x dx
0
e− x e− x
= E ( x) = x − 2
− 0
1
= E ( x ) = ( 0 − 0 ) − 0 − 2
1
Mean = E ( x ) =
Variance:
The variance of the exponential distribution is 2 = x f ( x ) dx −
2 2
−
2
1
2 = x 2 e− x dx −
0
e− x e− x e− x
= x2
2
− 2 x 2 + 2 3
− − 0
1
2 = ( 0 − 0 + 0 ) − 0 − 0 + 2 3
−
1
2 =
2
DSSM Compiled by: Magisetty Harikrushna 141
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
The cumulative distribution function
Let X be exponential distribution, then cumulative distribution function is F ( x ) = P ( X x ) .
When x 0
x
F ( x) = P ( X x) = f ( x ) dx = 0
−
When x 0
x
F ( x) = P ( X x) = f ( x ) dx
−
0 x
F ( x) = f ( x ) dx + f ( x ) dx
− 0
x
F ( x ) = 0 + e− x dx
0
x
F ( x ) = e− x dx
0
x
e− x
F ( x) =
− 0
F ( x ) = − ( e − x − e0 )
F ( x ) = 1 − e− x
0 x0
The cumulative distribution function is F ( x ) = − x
1 − e x0
142 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Let X be a random variable with the probability density function
−x
a e 3
f ( x) = x 0 Find (i) The value of ‘a’ (ii) P ( X 3) (iii) P (1 X 5)
0 else where
−x
a e 3
Solution: Given the probability density function f ( x ) = x0
0 else where
(i) We have f ( x ) dx = 1
−
0
f ( x ) dx + f ( x ) dx = 1
− 0
−x
0 + a e 3 dx = 1
0
−x
a e 3
dx = 1
0
−x
e 3
a =1
1
−
3 0
−3a ( e− − eo ) = 1
−3a ( 0 − 1) = 1
3a = 1
1
a=
3
1 −3x
e x0
The probability density function f ( x ) = 3
0
else where
(ii) P ( X 3)
P ( X 3) = f ( x ) dx
3
DSSM Compiled by: Magisetty Harikrushna 143
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1 −x
P ( X 3) = e 3 dx
3
3
−x
1
P ( X 3) = e 3 dx
33
−x
1 e 3
P ( X 3) =
3 −1
3 3
P ( X 3) = − ( e− − e−1 )
1
P ( X 3) = − 0 −
e
1
P ( X 3) =
e
(iii) P (1 X 5)
5
P (1 X 5) = f ( x ) dx
1
1 −x
5
P (1 X 5) = e 3 dx
1
3
5 −x
1
P (1 X 5) = e 3
dx
31
5
−x
1e 3
P (1 X 5 ) =
3 −1
3 1
−5 −1
P (1 X 5 ) = − e 3 − e 3
−1 −5
P (1 X 5 ) = e − e 3 3
144 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The time (in hours) required to repair a machine is exponential distributed with parameter
1
. What is the probability that the repair time exceeds 3 hours.
3
1
Solution: Given parameter =
3
e − x x0
The probability distribution is f ( x ) =
0 else where
1 − 3x
e x0
i.e., f ( x ) = 3
0
else where
The probability that the repair time exceeds 3 hours is P ( X 3)
P ( X 3) = f ( x ) dx
3
1 −3x
P ( X 3) = e dx
3
3
−x
1
P ( X 3) =
3 3
3
e dx
−x
1 e 3
P ( X 3) =
3 −1
3 3
P ( X 3) = − ( e− − e−1 )
1
P ( X 3) = − 0 −
e
1
P ( X 3) =
e
DSSM Compiled by: Magisetty Harikrushna 145
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: The length of time a person speaks over phone follows exponential distribution with mean
6. What is the probability that the person will take for (i) More than 8 minutes.
(ii) Between 4 and 8 minutes.
1
Solution: Given mean = =6
1
=
6
e − x x0
The probability distribution is f ( x ) =
0 else where
1 − 6x
e x0
i.e., f ( x ) = 6
0
else where
(i) More than 8 minutes
i.e., P ( X 8) = f ( x ) dx
8
1 −x
P ( X 8) = e 6 dx
8
6
−x
1
P ( X 8) = e 6 dx
68
−x
1e 6
P ( X 8) =
6−1
6 8
− − 86
P ( X 8) = − e − e
−
4
P ( X 8) = − 0 − e 3
4
−
P ( X 3) = e 3
146 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
(ii) Between 4 and 8 minutes.
8
P ( 4 X 8) = f ( x ) dx
4
1 −x
8
P ( 4 X 8) = e 6 dx
4
6
8−x
1
P ( 4 X 8) = e 6
dx
64
8
−x
1e 6
P ( 4 X 8) =
6−1
6 4
−8 −4
P ( 4 X 8) = − e 6 − e 6
−2 −4
P ( 4 X 8) = e − e 3 3
Problem: Assume that the time between arrivals of customers at a particular bank is exponentially
distributed with a mean of 4 minutes.
(i) Find the probability that the time between arrivals is greater than 5 minutes.
(ii) Find the probability that the time between arrivals is between 1 and 4 minutes.
DSSM Compiled by: Magisetty Harikrushna 147
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Deterministic & Stochastic Statistical Methods
Unit – III
Stochastic Processes and Markov Chains
Transition probability Matrix
A Transition Matrix, also, known as a stochastic or probability matrix is a square
matrix of order n n representing the transition probabilities of a stochastic system. The size
n of the matrix is linked to the cardinality of the State Space that describes the system being
modelled.
i.e., The matrix describing the Markov chain is called transition matrix.
1 2 … n
p11 p12 ... p1n
1
2 p21 p22 ... p2 n
P= ,
… ... ... ... ...
n pn1 pn 2 ... pnn
This matrix is said to be Transition, if
(i) It must be square matrix.
(ii) pij 0 for all i and j.
(iii) p ij = 1 (Row wise)
Note: In transition probability matrix,
Rows always represent NOW/FROM/PRESENT
Columns always represent NEXT/TO/FUTURE
Entry ( i, j ) is the probability of going From state ‘i’ To state ‘j’.
i.e., pij ( n ) = P X n +1 = j | X 1 = i
Row wise movement is retention and loss, Colum wise move is retention and gain, and
diagonal move is retention.
DSSM Compiled by: Magisetty Harikrushna 1
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Three boys A, B and C are throwing a ball at each other. A always throws the ball
to B, and B always throws the ball to C, but C is as likely to throw the ball to B as to A. Find
the transition probability matrix.
Solution: Clearly there are three states namely boys A, B and C.
So, the transition probability matrix is of order 3 3.
Given, A always throws the ball to B, B always throws the ball to C, but C is as likely to throw
the ball to B as to A.
From / To Boy A Boy B Boy C
Boy A 0 1 0
Boy B 0 0 1
Boy C ½ ½ 0
0 1 0
Therefore, the transition probability matrix is P = 0 0 1 .
1 1 0
2 2
2 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A professor of Statistics does not like to guess and decides on an innovative way of
assigning homework based on probabilities. On the first day of the week, he draws a transition
diagram as given below. The nodes of the diagram represent full-credit (F), half-credit (H) and
no-credit (N) assignments. The transition probabilities for day 1 are as shown in the diagram
construct TPM.
0.40
H
0.45
0.30 F N 0.10
0.65
Solution: Here we have three states namely full credit (F), half-credit (H) and no credit (N).
So, the transition probability matrix is of order 3 3.
From / To Full Credit (F) Half Credit (H) No Credit (N)
Full Credit (F) 0.30 0.25 0.45
Half Credit (H) 0.45 0.40 0.15
No Credit (N) 0.65 0.25 0.10
0.30 0.25 0.45
Therefore, the transition probability matrix is P = 0.45 0.40 0.15 .
0.65 0.25 0.10
DSSM Compiled by: Magisetty Harikrushna 3
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Markov process: A Markov chain or Markov process is a random process in which the
occurrence of future (Next) state depends on the immediately preceding (Current) state and
only on it, is known as Markov chain or Markov process.
Is a sequence of possibly dependent random variables x1 , x2 , x3 ,...xn identified by
increasing values of a parameter, mainly time with the property that any prediction of the next
value of the sequence xn , knowing the preceding states x1 , x2 , x3 ,...xn −1 , may be based on the
last state xn −1 alone. That is, the future value of such a variable is independent of its past history.
4 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A certain market survey is made on two brands of breakfast foods A and B. Every
time a customer purchases, the same brand or switches to another brand. The transmission
probabilities are given below, draw the Markov chain.
From/To Brand A Brand B
Brand A 0.8 0.2
Brand B 0.6 0.4
Solution: Given
From/To Brand A Brand B
Brand A 0.8 0.2
Brand B 0.6 0.4
0.4
The Markov process is
0.8 A B 0.6
0.2
DSSM Compiled by: Magisetty Harikrushna 5
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Classification of States in a Markov Chain
Path: Let ‘i’ and ‘j’ be two states in the chain, a path from i to j is a sequence of transitions
that begin in i and end in j, such that each transition in the sequence has a positive probability
of occurring.
Eg: In the given Markov chain
Here there is a path between state 1 and state 3, as we can
find a sequence of transitions that can take from state 1
to state 3, i.e., state 1 to state 2 and then state 2 to state 3
Reachable: A state j is reachable from state i, if there is a path leading form i to j.
Eg: In the given Markov chain
Here there is a path from that can take us from state A to state C,
So, we say that the state C is reachable from state A.
Communicate: Two states i and j are said to be communicate if j is reachable from i, and i is
reachable form j. Is denoted by i j 0.40
Eg: In the given Markov chain every state is communicate with other, 1
Since state 2 is reachable from 1 and 1 is reachable from 2,
state 3 is reachable from 2 and 2 is reachable from 3,
2 0.45 3
state 3 is reachable from 1 and 1 is reachable from 3.
0.65
6 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Communicability: A State is said to be communicability if it is
Reflexive: Any state communicates with itself
i.e., i i
Symmetric: If state i communicates with state j, then state j communicates with state i.
i.e., i j and j i
Transitive: If state i communicates with state j, and state j communicates with state k, then
state i communicates with state k
i.e., i j and j k then i k .
Communicative class: Two states that communicate with each other are in the same class. A
state that communicates with no other states itself is a class.
i.e., C ( i ) = j S / i j
Two states are said to be in the same class if the two states communicate with each other, that
is i ↔ j, then i and j are in same class.
Thus, all states in a Markov chain can be partitioned into disjoint classes
If states i and j are in the same class, then i ↔ j.
1
If a state i is in one class and state j is in another class,
then i and j do not communicate.
2 3
Here C (1) = 1, 2,3 , C ( 2 ) = 1,3 and C ( 3) = 1, 2
Closed Set: A set of states S in a Markov chain is a closed set if no state outside of S is
reachable from any state in S.
DSSM Compiled by: Magisetty Harikrushna 7
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Here the states 1, 2, 3 form a closed set and 4, 5 form a closed set. As the states 4 and 5 not
reachable from any other states 1, 2, 3.
The closed sets are 1, 2,3 and 4,5 .
Absorbing State: A state ‘i’ is an absorbing state if pii = 1 , an absorbing state is also a closed
set containing only one state.
Whenever we enter in absorbing state, we never leave the state.
Eg: Suppose a person has 1 coin in a game, he is going to gain a coin or lose a coin until that
person loses everything or gets 4 coins then that person quits the game.
If he gets 0 coins, he quits the game and if he gets 4 coins, he quits the game.
If the person has 1 coin, he may get zero coin if he loses the game, 0.5 0.5 0.5 0.5
or he may get 2 coins if he wins the game, then with 2 coins,
1 0 1 2 3 4 1
if he wins, he may get 3 coins or if he loses, he may get 1 coin,
next with 3 coins if he wins, he may get 4 coins then that person quits 0.5 0.5
the game or lose he may get 2 coins.
In the given Markov chain, the states 1 and 5 are absorbing states, since p11 = 1, p55 = 1
Transient: A state in a Markov chain is transient if it is not possible to return to that state in a
finite number of steps.
i.e., A state i is transient state if there exists a state j that is reachable from i, but the state i is
not reachable from j.
Eg: 0.5 0.5 0.5 0.5
1 1 2 3 4 5 1
0.5 0.5
In the Markov chain, we can go from state 2 to state 1 or 5, but not from 1 to 2 or 5 to 2,
Similarly,
we can go from state 3 to 1 or 5, but not from 1 to 3 or 5 to 3,
we can go from state 4 to 1 or 5, but not from 1 to 4 or 5 to 4.
Hence the states 1 and 5 are transient.
8 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Recurrent: A state in a Markov chain is recurrent if it is possible to return to that state in a
finite number of steps. 0.5 0.5
i.e., A state is recurrent if it is not transient.
1 2 3
In the above Markov chain, the states 1, 2 and 3 are recurrent.
0.5 0.5
Periodic: A state i is periodic with period k 1, if k is the smallest number such that all paths
leading from state i back to state i have a length that is a multiple of k.
A state which is not periodic is called aperiodic.
Note: If there is any self-transition, then it always aperiodic.
Eg: In diagram A the length of the shortest path that can
1 2 3
Take us from out of the state 1 and bring us back,
Here the length of that path is LS = k = 2 i.e., 1 to 2 and 2 to 1. Diagram A
The length of the other paths that can take us out of 1 and let us to go back to 1
is LO = 4 , i.e., 1 to 2, 2 to 3, and 3 to 2, 2 to 1.
Clearly the length of other path is multiple of shortest path, so the state 1 is periodic.
In diagram B the length of the shortest path that can take us from out of the state 1 and bring
us back, here the length of that path is LS = k = 2 i.e., 1 to 2 and 2 to 1.
The length of the other paths that can take us out of 1 and let us to go back to 1 1 2
is LO = 3 , i.e., 1 to 2, 2 to 3, and 3 to 1.
Here the length of another path is not multiple of shortest path, 3
Diagram B
so, the state 1 is aperiodic.
Irreducible: In Markov chain it is possible to transition from any state in the chain to any other
state, possibly with positive probability, in a finite number of steps.
Ergodicity: A Markov chain is ergodic if all states are irreducible, aperiodic and communicate
with each other.
DSSM Compiled by: Magisetty Harikrushna 9
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Reversibility: A Markov chain is reversible if the probability of transitioning from one state
to another is equal to the probability of transitioning from that state back to the original state.
Regular chain: A Markov chain is called a regular chain if some power of the transition matrix
has only positive elements. i.e., strictly greater than zero.
Note: There is a theorem that says that if an n n transition matrix represents n states, then we
need only examine powers T m up to m = ( n − 1) + 1 .
2
10 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Determine whether the following Markov chains are regular.
0.6 0 0.4 0.2 0.4 0.4
0 1 1 0
a) b) c) 0.2 0.4 0.4 d) 0.6 0.4 0
0.4 0.6 3 7 0 1 0.3 0.2 0.5
0
0.6 0 0.4
Solution: Let the given matrix be A = 0.2 0.4 0.4
0 1
0
Here the given matrix is of order 3 3 , so we need only examine powers up to ( n − 1) + 1 = 5.
2
0.36 0 0.64
A = 0.2 0.16 0.64 , …
2
0 1
0
0.07776 0 0.92224
A = 0.06752 0.01024 0.9224
5
0 1
0
Here all the entries of the 5th order matrix are not all positive, so the matrix is nor regular.
DSSM Compiled by: Magisetty Harikrushna 11
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Verify the following chains, Ergodic or not.
0.5 0.5 0 0
0.4 0.6 0 0.25 0.5 0.25
(i) 0.5
0.5 0.5 0 0
0 0.5 (ii) (iii) 0.6 0.4 0
0 0.25 0.75 0 0 0.6 0.4 0
0.6 0.4
0 0 0.25 0.75
0 0.8 0.2
(iv) 0.3 0.7 0
0.4 0.5 0.1
Solution: (i) The Markov chain is
1
Here all the states are recurrent, each state communicate with other state,
2
there is a self-transition for state 1 and state 3, so the states are aperiodic.
Therefore, the chain is Ergodic. 3
(ii) The Markov chain is
1 2
Here all the states are recurrent, but states 1and 2 not communicate with
States 3 and 4. So states are not communicated. there is a self-transition
3 4
for each state, so the states are aperiodic.
Therefore, the chain is not Ergodic.
(iii) The Markov chain is
1 2
Here all the states are recurrent, each state communicate with other state,
there is a self-transition for states 1, 2 and 3, so the states are aperiodic.
3
Therefore, the chain is Ergodic.
(iv) The Markov chain is
3 1 other state,
Here all the states are recurrent, each state communicate with 1
there is a self-transition for states 1, 2 and 3, so the states are aperiodic.
Therefore, the chain is Ergodic.
2 3
12 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Conditional Probability of the Markov Chain
Let X 0 , X1 , X 2 ,... be a Markov chain with n n transition probability matrix P. Then the
t–step transition probabilities are given by the matrix P t .
i.e., P ( X t = j / X 0 = i ) = pijt = P ( X n+t = j / X n = i ) , where pijt is the element of the matrix P t
Transition probability
One step time period
The probability of moving from one state of a system into another state is the transition
probability. In a Markov chain Pij is the transition probability of going from ith state to jth
state at the one step time period.
i.e., pij = P X n +1 = j | X n = i
If P X 2 = 2 | X 1 = 1 = pij = p12 ;
P X 2 = 3 | X1 = 2 = pij = p23 are the one step transition probabilities.
2- step time period
Probability form state ‘i’ to state ‘j’ after 2- step time period, denoted by pij (2) is defined as
pij (2) = P X 3 = j | X 1 = i
If P X 2 = 2 | X 0 = 3 = p32
(2)
;
P X 3 = 3 | X1 = 2 = p23
(2)
are the two step transition probabilities.
n- step time period
Probability form state ‘i’ to state ‘j’ after n- step time period, denoted by pij ( n ) is defined as
pij ( n ) = P X n +1 = j | X 1 = i
Note: In the Markov process, two terms States and Transition Probability play a key role.
DSSM Compiled by: Magisetty Harikrushna 13
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Notation:
S o is initial probability of the states.
S 1 is probability of the states after the 1st time period.
S 2 is probability of the states after the 2nd time period.
…
S n is probability of the states after the nth time period.
P is TPM after the 1st time period.
P2 is TPM after the 2nd time period.
…
P n is TPM after the nth time period.
Note: X n = i ; P X n a Sn a
So, we calculate the probability of states by using the formula S n S 0 P n or S n 1 SnP
14 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A man travels to work daily, alternating between driving his car and taking the train.
He never takes the train two consecutive days. If he drives one day, the next day, he's equally
likely to drive or take the train. Determine the TPM. Also, calculate the probability that he
switches from taking the train to driving exactly four days later.
T C
T 0 1
Solution: The transition probability matrix is P = 1 1
C
2 2
The probability that he changes from going by train to driving exactly in four days is given by
P ( X 4 = C / X 0 = T ) = pTC
4
P ( X n+t = j / X n = i ) = pijt
0.375 0.625
P4 =
0.3125 0.6875
P ( X 4 = C / X 0 = T ) = pTC
4
= 0.625
DSSM Compiled by: Magisetty Harikrushna 15
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A company signs each year a contract for the delivery of a certain product with one
of three possible subcontractors A, B and C. If company signs subcontractor A, there is 60%
chance that would continue with subcontractor A, 30% and 10% chance that would shift to
subcontractors B and C, respectively. If company signs subcontractor B, there is 40% chance
that would shift to subcontractor A and 20% chance to subcontractor C. If the company signs
subcontractor C, there is 50%, 30% chance that would shift to subcontractors A and B
respectively. Using this information,
(i) write the transition probability matrix.
(ii) Find the probability that subcontract is switched from A to B after two years period.
(ii) Find the probability that subcontract is switched from B to C after two years period.
(iii) Find the probability that subcontract C will be able to retain its after two years period.
0.6 0.3 0.1
Solution: The transition probability matrix to the given data is P = 0.4 0.4 0.2
0.5 0.3 0.2
The probability that subcontract from A will be shifted to subcontract B after two years period
is given by P ( X 2 = B / X 0 = A) = pAB
2
0.53 0.33 0.14
Here P = 0.5 0.34 0.16
2
0.52 0.33 0.15
P ( X 2 = B / X 0 = A) = p AB
2
= 0.33
There is 33% chance that subcontract from A will be shifted to subcontract B after two years
period.
The probability that subcontract from B will be shifted to subcontract C after two years period
is given by P ( X 2 = C / X 0 = B ) = pBC
2
P ( X 2 = B / X 0 = A) = pBC
2
= 0.16
16 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
There is 16% chance that subcontract from B will be shifted to subcontract C after two years
period.
The probability that subcontracts C will be able to retain its after two years period is given by
P ( X 2 = C / X 0 = C ) = pCC
2
P ( X 2 = C / X 0 = C ) = pCC
2
= 0.15
There is 15% chance that subcontract C will be able to retain its after two years period.
DSSM Compiled by: Magisetty Harikrushna 17
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Three children (denoted by 1, 2, 3) arranged in a circle play a game of throwing a
ball to one another. At each stage the child having the ball is equally likely to throw it into any
one of the other two children. Suppose that X o denotes the child who had the ball initially and
X n , ( n 1) denotes the child who had the ball after n throws. Determine the transition
probability matrix P. Calculate P ( X 2 = 1/ X 0 = 1) , P ( X 2 = 2 / X 0 = 3) , P ( X 2 = 3 / X 0 = 2 )
and the probability that the child who had originally the ball will have it after 2 throws.
Solution: Given that at each stage the child having the ball is equally likely to throw it into
any one of the other two children.
The transition probability matrix P is
To
1 2 3
1 40 0.5 0.5
From 2 0.5 0 0.5
3 0.5 0.5 0
1 2 3
1 4 0.5 0.25 0.25
P = 2 0.25 0.5 0.25
2
3 0.25 0.25 0.5
We have P X n+1 = j | X1 = i = pij ( n )
P ( X 2 = 1/ X 0 = 1) = p112
P ( X 2 = 1/ X 0 = 1) = p112 = 0.5
P ( X 2 = 2 / X 0 = 3) = p32
2
P ( X 2 = 2 / X 0 = 3) = p32
2
= 0.25
P ( X 2 = 3 / X 0 = 2 ) = p23
2
P ( X 2 = 3 / X 0 = 2 ) = p23
2
= 0.25
If the child 1 has the ball at initially, then the initial probability is S 0 1 0 0
After two throws the probability is S2 = S0 P2 S n S 0P n
18 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0.5 0.25 0.25
S2 = (1 0 0 ) 0.25 0.5 0.25 = ( 0.5 0.25 0.25 )
0.25 0.25 0.5
After two throws
Probability of child 1 having the ball is 50%,
Probability of child 2 having the ball is 25%
Probability of child 3 having the ball is 25%
If the child 2 has the ball at initially, then the initial probability is S 0 0 1 0
After two throws the probability is S2 = S0 P2 S n S 0P n
0.5 0.25 0.25
S2 = ( 0 1 0 ) 0.25 0.5 0.25 = ( 0.25 0.5 0.25 )
0.25 0.25 0.5
After two throws
Probability of child 1 having the ball is 25%,
Probability of child 2 having the ball is 50%
Probability of child 3 having the ball is 25%
If the child 3 has the ball at initially, then the initial probability is S 0 0 0 1
After two throws the probability is S2 = S0 P2 S n S 0P n
0.5 0.25 0.25
S2 = ( 0 0 1) 0.25 0.5 0.25 = ( 0.25 0.25 0.5 )
0.25 0.25 0.5
After two throws
Probability of child 1 having the ball is 25%,
Probability of child 2 having the ball is 25%
Probability of child 3 having the ball is 50%
DSSM Compiled by: Magisetty Harikrushna 19
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: In a classroom 80% of the students who do Mathematics work during one study
period will do the Mathematics work at the next study period. 30% of the students who do
Physics work during one study period will do the Physics work at the next study period. Initially
there were 60 students doing Mathematics and 40 students doing Physics work.
(i) Write the transition probability matrix.
(ii) Calculate the number of students who do mathematics work and Physics work for the next
subsequent two study periods.
0.8 0.2
Solution: The transition probability matrix is P =
0.7 0.3
Given the initial probability is S 0 60 40
0.78 0.22
P2 =
0.77 0.23
We have S2 = S0 P2 S n S 0P n
0.78 0.22
S2 = ( 60 40 )
0.77 0.23
S2 = ( 77.6 22.4 )
After two study periods there will be approximately 78 students do Mathematics and 22
students do Physics.
20 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: The price of an equity share of a company may increase, decrease or remain constant
on any given day. It is assumed that the change in price on any day affects the change on the
following day as decreased by the following transition matrix
Tomorrow
Inc. Dec. Unch.
0.5 0.2 0.3
Inc.
Today Dec. 0.7 0.1 0.2
Unch. 0.4 0.5 0.1
(i) If the price of share increased today, what are chances that it will increase, decrease or
remain unchanged tomorrow?
(ii) If the prices of the share decreased today, what are the chances that it will increase
tomorrow?
(iii) If the price of the share remained unchanged today, what are the chances that it will
increase, decrease or remain unchanged day after tomorrow?
Solution: Given transition probability matrix is
Inc. Dec. Unch.
0.5 0.2 0.3
Inc.
Today Dec. 0.7 0.1 0.2
Unch. 0.4 0.5 0.1
(i) If the price of share increased today, then the initial probability is S 0 1 0 0
Now S1 = S0 P S n S 0P n
0.5 0.2 0.3
S1 = (1 0 0 ) 0.7 0.1 0.2 = ( 0.5 0.2 0.3)
0.4 0.5 0.1
Thus, the chance that the price will increase, decrease or remain unchanged tomorrow are 50%,
20% and 30%.
(ii) If the prices of the share decreased today then the initial probability is S 0 0 1 0
Now S1 = S0 P S n S 0P n
DSSM Compiled by: Magisetty Harikrushna 21
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
0.5 0.2 0.3
S1 = ( 0 1 0 ) 0.7 0.1 0.2 = ( 0.7 0.1 0.2 )
0.4 0.5 0.1
Thus, the chance that the price will increase tomorrow is 70%.
(iii) If the price of the share remained unchanged today, then initial probability is
S0 0 0 1
Tomorrow will be S1 = S0 P S n S 0P n
0.5 0.2 0.3
S1 = ( 0 0 1) 0.7 0.1 0.2 = ( 0.4 0.5 0.1)
0.4 0.5 0.1
Day after tomorrow will be S2 = S0 P2
0.51 0.27 0.22
S2 = ( 0 0 1) 0.5 0.25 0.25 = ( 0.59 0.18 0.23)
0.59 0.18 0.23
Thus, the chance that the price will increase, decrease or remain unchanged day after tomorrow
are 59%, 18% and 23%.
22 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Three manufacturers X, Y and Z are competing with each other. The following
matrix gives the transition probabilities that customers will move from one manufacturer to
the other in any month. Interpret the matrix in terms of (a) retention and loss. (b) retention
and gain.
To
X Y Z
X 40.7 0.1 0.2
From Y 0.1 0.8 0.1
Z 0.2 0.1 0.7
Solution: In the given matrix each element value between 0 and 1 and the sum of elements in
each row is exactly one. Thus, the given matrix is a transition matrix.
(a) Each row of the matrix can be used to interpret the monthly retention and loss by each
manufacturer.
Row 1: According to the data of row 1, in the coming month manufacturer X will retain 70%
of its customers but losses 10% to B and 20% to C.
i.e., Given that a customer purchased manufacturer X's product in the past month, the
probability of a repeat purchase of that product from manufacturer X is 0.7, and the probability
of a switch to manufacturer Y's product is 0.1 and manufacturer Zs product is 0.2.
Row 2: According to the data of row 2, in the coming month manufacturer Y will lose 10% of
its previous month's customers to manufacturer X and retain 80% of them itself and will lose
10% to Z.
Row 3: According to the data of row 3, in the coming month manufacturer Z will retain 70%
of its customers but loses 20% to X and 10% to Z.
(b) Similarly, columns can be used to study the retention and gains of each manufacturer.
Column 1: The data of column 1, reveals that manufacturer X will retain 70% of last month's
customers and also gains 10% customers of B and 20% customers of C.
Column 2: The data of column 2 reveals that manufacturer Y will gain 10% of X's customers,
10% of Zs customers and retain 80% of its own.
Column 3: The data of column ,3 reveals that manufacturer Z will gain 20% of X's customers,
10% of Y's customers and retain 70% of its own.
DSSM Compiled by: Magisetty Harikrushna 23
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Mr. Arjun purchases a new car every two years. He prefers a brand A car but
sometimes buys another brand. For simplicity, suppose we designate as brand B any other car
which he may purchase. To add a little spice to the car buying Mr. Arjun has devised a method
of randomly selecting each new car. He places ten slips numbered 1 through 10, in a bowl and
draws one slip. If the present car is brand A and the number on the slip is 8 or less, he will
again purchase brand A. Thus, there is a probability of 0.8 that the next car will also be brand
A. If he presently owns a brand B car and the number on the slip is 6 or less, he will next
purchase a brand A car. The probability they switch back to a brand A car is then 0.6.
(i) Construct and interpret the state transition matrix in terms of (a) retention and loss.
(b) retention and gain.
(ii) Calculate the probability of Mr. Arjun purchasing brand A car at the end of second period,
third period. Draw the transition probability diagrams and the transition trees.
(iii) Calculate the probability of Mr. Arjun purchasing brand B car at the end of second period,
third period.
Solution:
(i) Let state 1 be the condition of owning a brand A car and state 2 that of owning a brand B
car. The transition probabilities are then p11 = 0.8, p12 = 0.2, p21 = 0.6 and p22 = 0.4 The state-
0.8 0.2
transition matrix P is P =
0.6 0.4
(a) Retention and Loss (Rows of P)
p11 = 0.8 means that Mr. Arjun now using brand A car will again purchase brand A car at the
next purchase in 8 out of 10 times. This implies retention to brand A car.
p12 = 0.2 means that Mr. Arjun now using brand A car will switch over to brand B car the next
purchase in 2 out of 10 times. This implies loss to brand A car.
similarly, we can interpret the second row of P.
24 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
(b) Retention and Gain (Columns of P).
p11 = 0.8 means that Mr. Arjun now using brand A car will again purchase brand A car at the
next purchase in 8 out of 10 times. This implies retention to brand A car.
p21 = 0.6 shows that Mr. Arjun now using brand B car will switch over to brand A car next
time in 6 out of 10 times. This implies gain to brand A car.
similarly, we can interpret the second column of P.
(ii) A transition probability diagram and Transition trees
0.6
0.8 A B 0.4
0.2
0.8
0.8
0.2
Current Purchase
0.6
0.2
0.4
p11 = ( 0.8)( 0.8) + ( 0.2 )( 0.6 ) = 0.76
p12 = ( 0.8)( 0.2 ) + ( 0.2 )( 0.4 ) = 0.24
0.8
0.6
0.2
Current Purchase
0.6
0.4
0.4
p21 = ( 0.6 )( 0.8) + ( 0.4 )( 0.6 ) = 0.72
p22 = ( 0.6 )( 0.2 ) + ( 0.4 )( 0.4 ) = 0.28
DSSM Compiled by: Magisetty Harikrushna 25
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
If Mr. Arjun starts by buying a brand A car then initial probability is S 0 1 0
After the first transition S1 = S0 P S n S 0P n
0.8 0.2
S1 = (1 0 ) = ( 0.8 0.2 )
0.6 0.4
After the second transition S2 = S0 P2 S n S 0P n
0.76 0.24
S2 = (1 0 ) = ( 0.76 0.24 )
0.72 0.28
After the third transition S3 = S0 P3 S n S 0P n
0.752 0.248
S3 = (1 0 ) = ( 0.752 0.248)
0.744 0.256
If Mr. Arjun starts by buying a brand B car then initial probability is S 0 0 1
After the first transition S1 = S0 P S n S 0P n
0.8 0.2
S1 = ( 0 1) = ( 0.6 0.4 )
0.6 0.4
After the second transition S2 = S0 P2 S n S 0P n
0.76 0.24
S2 = ( 0 1) = ( 0.72 0.28)
0.72 0.28
After the third transition S3 = S0 P3 S n S 0P n
0.752 0.248
S3 = ( 0 1) = ( 0.744 0.256 )
0.744 0.256
26 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Consider a bike share problem with only 3 stations: A, B, C. Suppose that all bikes must
be returned to the station at the end of the day, so that all the bikes are at some station. Each day,
the distribution of bikes at each station changes, as the bikes get returned to different stations from
where they are borrowed. Of the bikes borrowed from station A, 30% are returned to station A,
50% end up at station B, and 20% end up at station C. Of the bikes borrowed from station B, 10%
end up at station A, 60% have been returned to station B, and 30% end up at station C. Of the bikes
borrowed from station C, 10% end up at station A, 10% end up at station B, and 80% are returned
to station C.
(i) Express this information as a transition probability matrix and determine the probabilities of
bike being at a particular station after two days.
(ii) Suppose when we start observing the bike share program, 30% of the bikes are at station A,
45% of the bikes are at station B and 25% are at station C, determine the distribution of bikes at
the end of the next day and after two days.
Solution:
(i) The given Markov chain is
0.3 0.5
0.1 0.6
0.8
0.3 0.5 0.2
The transition probability matrix is P = 0.1 0.6 0.3
0.1 0.1 0.8
0.16 0.47 0.37
P = 0.12 0.44 0.44
2
0.12 0.19 0.69
After two days:
If today the bike is at station A then the initial probability is S0 = (1 0 0 )
DSSM Compiled by: Magisetty Harikrushna 27
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
We have S 2 S 0 P 2 S n S 0P n
0.16 0.47 0.37
S2 = (1 0 0 ) 0.12 0.44 0.44 = ( 0.16 0.47 0.37 )
0.12 0.19 0.69
Thus, after two days, the chance that the bike will be at station A is 16%, at station B is 47%
and at station C is 37%.
If today the bike is at station B then the initial probability is S0 = ( 0 1 0 )
We have S 2 S 0 P 2 S n S 0P n
0.16 0.47 0.37
S2 = ( 0 1 0 ) 0.12 0.44 0.44 = ( 0.12 0.44 0.44 )
0.12 0.19 0.69
Thus, after two days, the chance that the bike will be at station B is 12%, at station B is 44%
and at station C is 44%.
If today the bike is at station C then the initial probability is S0 = ( 0 0 1)
We have S 2 S 0 P 2 S n S 0P n
0.16 0.47 0.37
S2 = ( 0 0 1) 0.12 0.44 0.44 = ( 0.12 0.19 0.69 )
0.12 0.19 0.69
Thus, after two days, the chance that the bike will be at station C is 12%, at station B is 19%
and at station C is 69%.
(ii) Given the initial probability S0 = ( 0.3 0.45 0.25 )
The distribution of bikes at the end of the next day
We have S 1 S 0 P S n S 0P n
0.3 0.5 0.2
S1 = ( 0.3 0.45 0.25) 0.1 0.6 0.3 = ( 0.160 0.445 0.395 )
0.1 0.1 0.8
At the end of the next day 16% of the bikes are at station A, 44.5% of the bikes are at station B
and 39.5% of the bikes are at station C.
28 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
The distribution of bikes at the end of after two days.
We have S 2 S 0 P 2 S n S 0P n
0.16 0.47 0.37
S2 = ( 0.3 0.45 0.25) 0.12 0.44 0.44 = ( 0.1320 0.3865 0.4815)
0.12 0.19 0.69
At the end of after two days 13.2% of the bikes are at station A, 38.65% of the bikes are at station
B and 48.15% of the bikes are at station C.
DSSM Compiled by: Magisetty Harikrushna 29
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A man either uses his car or takes a bus or a train to work each day. The TPM of
0.1 0.5 0.4
the Markov chain with these three states is P = 0.6 0.2 0.2 and the initial probability is
0.3 0.4 0.3
( 0.7 0.2 0.1) calculate (i) P ( X 2 = 1) (ii) P ( X 2 = 3)
0.1 0.5 0.4
Solution: Given TPM is P = 0.6 0.2 0.2
0.3 0.4 0.3
And initial probability is S0 = ( 0.7 0.2 0.1)
This means that 70% chance that a person uses car to go to work at the first day.
20% chance that a person uses bus to go to work at the first day.
10% chance that a person uses train to go to work at the first day.
0.1 0.5 0.4
Now P = 0.6 0.2 0.2
0.3 0.4 0.3
0.43 0.31 0.26
Implies P = 0.24 0.42 0.34
2
0.36 0.35 0.29
Now S 2 S 0 P 2 S n S 0P n
0.43 0.31 0.26
S2 = ( 0.7 0.2 0.1) 0.24 0.42 0.34
0.36 0.35 0.29
S2 = ( 0.385 0.336 0.279 )
(i) We have to find P ( X 2 = 1)
i.e., The probability that on the second day a man uses car to go to work.
P ( X 2 = 1) = S2 (1) P ( X n = a ) = Sn ( a )
P ( X 2 = 1) = S2 (1) = 0.385 or 38.5%
30 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
(ii) We have to find P ( X 2 = 3)
i.e., The probability that on the second day a man uses train to go to work.
P ( X 2 = 3) = S 2 ( 3) P ( X n = a ) = Sn ( a )
P ( X 2 = 3) = S2 ( 3) = 0.279 or 27.9%
DSSM Compiled by: Magisetty Harikrushna 31
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A market research team has surveyed customer buying habits concerning three brands
of talcum power in an area. It estimates at present, 20% of the customers buy brand A, 50% of the
customers buy brand B and 30% of the customers buy brand C. In addition, the team has analysed
its survey and has determined the following brand switching matrix.
Brand Next Bought
A B C
Brand Just Bought
A 0.6 0.3 0.1
B 0.4 0.5 0.1
C 0.2 0.1 0.7
Determine the expected distribution of consumers two time periods later.
0.6 0.3 0.1
Solution: The transition probability matrix is P = 0.4 0.5 0.1
0.2 0.1 0.7
The initial probability is S0 = ( 0.2 0.5 0.3)
0.5 0.34 0.16
P = 0.46 0.38 0.16
2
0.3 0.18 0.52
We have S 2 S 0 P 2 S n S 0P n
0.5 0.34 0.16
S2 = ( 0.2 0.5 0.3) 0.46 0.38 0.16 = ( 0.420 0.312 0.268)
0.3 0.18 0.52
S2 = ( 0.420 0.312 0.268)
P ( X 2 = A) = S2 ( A) = 0.420
P ( X 2 = B ) = S2 ( B ) = 0.312
P ( X 2 = C ) = S2 ( C ) = 0.268
1 1 1
The expected distribution of consumers two time periods later is .
0.420 0.312 0.268
32 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0.2 0.5 0.3
Problem: The transition probability matrix 0.5 0.3 0.2 is given with three states 1, 2 and 3.
0.3 0.3 0.4
And initial probability is ( 0.6 0.3 0.1) then evaluate (i) P ( X 2 = 1) and P ( X 2 = 3)
(ii) P ( X 3 = 2, X 2 = 3, X1 = 3, X 0 = 2 )
0.2 0.5 0.3
Solution: Given TPM is P = 0.5 0.3 0.2
0.3 0.3 0.4
And initial probability is S0 = ( 0.6 0.3 0.1)
0.38 0.34 0.28
Implies P = 0.31 0.4 0.29
2
0.33 0.36 0.31
We have S 2 S 0 P 2 S n S 0P n
0.38 0.34 0.28
S2 = ( 0.6 0.3 0.1) 0.31 0.4 0.29
0.33 0.36 0.31
S2 = ( 0.354 0.36 0.286 )
(i) P ( X 2 = 1) and P ( X 2 = 3)
P ( X 2 = 1) = S2 (1) = 0.354
P ( X 2 = 3) = S2 ( 3) = 0.286
(ii) P ( X 3 = 2, X 2 = 3, X1 = 3, X 0 = 2 )
P ( X 3 = 2, X 2 = 3, X1 = 3, X 0 = 2 ) = S0 ( 2 ) p23 p33 p32 1 1 1
2 3 3 2
= 0.3 0.2 0.4 0.3 = 0.0072
P ( X 3 = 2, X 2 = 3, X1 = 3, X 0 = 2 ) = 0.018
0.38 0.34 0.28
Now P = 0.31 0.4 0.29
2
0.33 0.36 0.31
DSSM Compiled by: Magisetty Harikrushna 33
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
We have S 2 S 0 P 2 S n S 0P n
0.38 0.34 0.28
S2 = ( 0.6 0.3 0.1) 0.31 0.4 0.29
0.33 0.36 0.31
S2 = ( 0.354 0.36 0.286 )
(i) P ( X 2 = 1) and P ( X 2 = 3)
P ( X 2 = 1) = S2 (1) = 0.354
P ( X 2 = 3) = S2 ( 3) = 0.286
34 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0.2 0.3 0.5
Problem: The transition probability matrix 0.1 0.6 0.3 is given with three states 0, 1 and 2.
0.4 0.3 0.3
And initial probability is ( 0.5 0.3 0.2 ) then evaluate (i) P ( X 3 = 2, X 2 = 1, X 1 = 0, X 0 = 2 )
(ii) P ( X 3 = 2, X1 = 0, X 0 = 2 ) (iii) P ( X 2 = 2 )
0 1 2
0 0.2 0.3 0.5
Solution: Given TPM is P = 1 0.1 0.6 0.3
2 0.4 0.3 0.3
And initial probability is S0 = ( 0.5 0.3 0.2 )
0.27 0.39 0.34
P = 0.20 0.48 0.32
2
1 1 1
2 0 1 2
0.23 0.39 0.38
(i) P ( X 3 = 2, X 2 = 1, X 1 = 0, X 0 = 2 )
Here we start from 2 move to 0, here time length is 1
then 0 to 1 with time length 1, finally 1 to 2 with time length 1.
So, P ( X 3 = 2, X 2 = 1, X 1 = 0, X 0 = 2 ) = S0 ( 2 ) p20 p01 p12
= 0.2 0.4 0.3 0.3 = 0.0072
P ( X 3 = 2, X 2 = 1, X1 = 0, X 0 = 2 ) = 0.0072
(ii) P ( X 3 = 2, X1 = 0, X 0 = 2 )
Here we start from 2 move to 0, here time length is 1
then 0 to 2 with time length 2. 1 2 2
2 0
So, P ( X 3 = 2, X1 = 0, X 0 = 2 ) = S0 ( 2 ) p20 p02
2
P ( X 3 = 2, X 2 = 1, X1 = 0, X 0 = 2 ) = 0.2 0.4 0.34 = 0.0272
(iii) P ( X 2 = 2 )
We have S 2 S 0 P 2 S n S 0P n
DSSM Compiled by: Magisetty Harikrushna 35
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
0.27 0.39 0.34
S2 = ( 0.5 0.3 0.2 ) 0.20 0.48 0.32
0.23 0.39 0.38
S2 = ( 0.241 0.417 0.342 )
P ( X 2 = 2 ) = S2 ( 2 ) = 0.342 P ( X n = a ) = Sn ( a )
36 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0.1 0.4 0.5
Problem: The TPM of the Markov chain with three states 1, 2 and 3 is 0.2 0.4 0.4
0.3 0.4 0.3
and initial probability is ( 0.6 0.3 0.1) then evaluate (i) P ( X 3 = 3, X 1 = 1) (ii) P ( X 2 = 3)
1 2 3
1 0.1 0.4 0.5
Solution: Given TPM is P = 2 0.2 0.4 0.4
3 0.3 0.4 0.3
And initial probability is S0 = ( 0.6 0.3 0.1)
0.24 0.4 0.36
P = 0.22 0.4 0.38
2
0.2 0.4 0.4
(i) P ( X 3 = 3, X 1 = 1)
Here we start from 1 move to 3, here time length is 2
So, P ( X 3 = 3, X 1 = 1) = S0 (1) p132
= 0.6 0.36 = 0.216
2 3
1
P ( X 3 = 3, X1 = 1) = 0.216
(ii) P ( X 2 = 3)
We have S 2 S 0 P 2 S n S 0P n
0.24 0.4 0.36
S2 = ( 0.6 0.3 0.1) 0.22 0.4 0.38
0.2 0.4 0.4
S2 = ( 0.23 0.4 0.37 )
P ( X 2 = 3) = S2 ( 3) = 0.37
DSSM Compiled by: Magisetty Harikrushna 37
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: The transition probability matrix of a Markov chain with states 0, 1, 2 is
34 14 0
P = 14 12 14 and the initial state distribution of the chain is P ( X 0 = i ) = 13 , i = 0,1,2 .
0 3 1
4 4
Obtain (i) P ( X 2 = 2 ) (ii) P ( X 2 = 2, X 1 = 1, X 0 = 2 )
34 1
4 0
1
Solution: Given the transition probability matrix P = 14 1
2 4
0 3 1
4 4
And initial probability is S0 = 13 ( 1
3
1
3 )
85 165 1
5 16
Now P 2 = 16 1 3
2 16
3 9 1
16 16 4
We have S 2 S 0 P 2 S n S 0P n
S 2 S0P 2
85 165 1
5 16
3
(
S2 = 13 1
3
1
)
3 16 2 16 = 8
1 3
( 11
24
1
6 )
3 9 1
16 16 4
(i) P ( X 2 = 2 )
P ( X 2 = 2 ) = S2 ( 2 ) = 11
24
(ii) P ( X 2 = 2, X 1 = 1, X 0 = 2 )
Here we start from 2 move to 1, here time length is 1
1 1 2
then 1 to 2 with time length 1. 2 1
So, P ( X 2 = 2, X1 = 1, X 0 = 2 ) = S0 ( 2 ) p21 p12
1 1 1 1
P ( X 2 = 2, X 1 = 1, X 0 = 2 ) = =
3 4 4 48
38 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Stationary Probability or Steady State Probability:
Let X n , n 0 be a homogeneous Markov chain with transition probability matrix P.
if there exists a probability vector S such that SP = S . Then S is called a stationary probability
or steady state probability or limiting distribution for the Markov chain.
Note: If the stationary distribution S exists then it is unique.
Problem: Given that a person last cola purchase was Coke, there is a 90% chance that his next
cola purchase will also be Coke. If a person last cola purchase was Pepsi, there is a 80% chance
that his next cola purchase will also be Pepsi. The present market share of the Coke and Pepsi
is 55% and 45% respectively. Construct the TPM. In the long run, what is the market
share of such cola?
0.90 0.10
Solution: The Transition Probability matrix is P =
0.20 0.80
The present share S0 = 0.55 0.45
Let S = p1 p2 with p1 + p2 = 1 be the stationary distribution such that SP = S
0.90 0.10
i.e., p1 p2 = p1 p2
0.20 0.80
implies 0.90 p1 + 0.20 p2 = p1 → (1)
0.10 p1 + 0.80 p2 = p2 → (2) and
p1 + p2 = 1 → (3)
2 1
By solving (1) and (3), we get p1 = , p2 =
3 3
Hence in the long run, the market share of the Coke and Pepsi will be 66.67% and 33.33%
respectively.
DSSM Compiled by: Magisetty Harikrushna 39
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: An auto insurance company classifies its customers in three categories: poor,
satisfactory and preferred. No one moves from poor to preferred or from preferred to poor in
one year. 40% of the customers in the poor category become satisfactory, 30% of those in the
satisfactory category moves to preferred, while 10% become poor; 20% of those in the
preferred category are downgraded to satisfactory.
(a) Write the transition matrix for the model.
(b) What is the limiting fraction of drivers in each of these categories?
Solution: The transition probabilities for this Markov chain with three states are as follows
Poor Satisfactory Preferred
Poor 0.6 0.4 0.0
Satisfactory 0.1 0.6 0.3
Preferred 0.0 0.2 0.8
0.6 0.4 0
The Transition probability matrix is 0.1 0.6 0.3
0 0.2 0.8
Let S = p1 p2 p3 with p1 + p2 + p3 = 1, be the limiting fraction of drivers such that SP = S
0.6 0.4 0
i.e., p1 p2 p3 0.1 0.6 0.3 = p1 p2 p3
0 0.2 0.8
implies 0.6 p1 + 0.4 p2 = p1
−0.4 p1 + 0.4 p2 = 0
p1 = p2 → (1)
0.1 p1 + 0.6 p2 + 0.3 p3 = p2
0.1 p1 − 0.4 p2 + 0.3 p3 = 0 → (2)
0.2 p2 + 0.8 p3 = p3
40 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0.2 p2 − 0.2 p3 = 0 → (3) and
p1 + p2 + p3 = 1 → (4)
1 4 6
By solving (1), (2) and (4), we get p1 = , p2 = , p3 =
11 11 11
1
Thus, the limiting fraction of drivers in the POOR category is , in the SATISFACTORY
11
4 6
category is , and in the PREFERRED category is .
11 11
DSSM Compiled by: Magisetty Harikrushna 41
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: In a certain market, there are three brands of Face creams A, B and C. Given that a
lady last purchased Face cream of brand A, there is 70% chance that she would continue with
brand A, 20% and 10% chances that she would shift to brands B and C, respectively. Given
that a lady last purchased Face cream of brand B, there is 50% chances that she would shift to
brand A and 10% chance to brand C. Given that a lady last purchased Face cream of brand C,
there is 60%, 20% chance that she would shift to brands A and B respectively. The present
market share of the three brands A, B and C is 60%, 30% and 10% respectively. Using this
information, find the market share of the brands A, B and C in the steady-state.
0.70 0.20 0.10
Solution: The Transition Probability matrix is P = 0.50 0.40 0.10
0.60 0.20 0.20
The present market share of the three brands is S0 = 0.60 0.30 0.10
Let S = p1 p2 p3 with p1 + p2 + p3 = 1, be the stationary distribution such that SP = S
0.70 0.20 0.10
i.e., p1 p2 p3 0.50 0.40 0.10 = p1 p2 p3
0.60 0.20 0.20
implies 0.70 p1 + 0.50 p2 + 0.60 p3 = p1
−0.30 p1 + 0.50 p2 + 0.60 p3 = 0 → (1)
0.20 p1 + 0.40 p2 + 0.20 p3 = p2
0.20 p1 − 0.60 p2 + 0.20 p3 = 0 → (2)
0.10 p1 + 0.10 p2 + 0.20 p3 = p3
0.10 p1 + 0.10 p2 − 0.80 p3 = 0 → (3)
p1 + p2 + p3 = 1 → (4)
By solving (1), (2) and (4), we get p1 = 0.63889, p2 = 0.25000, p3 = 0.11111
42 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Hence in the steady-state, the three brands’ shares A, B and C is will be 63.89%, 25% and
11.11% respectively.
Problem: A hospital operates on a charity basis. All expenses are paid by the government. Of
late, the board of governors of the hospital has been complaining about the size of the budget
and insisting that the hospital cut expenses. The major area of concern has been cost of keeping
patients in the intensive care unit (ICU). The cost has averaged Rs. 1000 per week per person
compared to only Rs. 500 per week per person for keeping patients in the WARDS.
Past history shows that of the patients in the ICU at the beginning of the week, 50% will remain
there at the end of the week, 50% will be moved to a WARD. Of the patients in the WARDS
at the beginning of the week, 50% will remain there at the end of the week, 10% will get worse
and be transferred to the ICU, and 40% will become OUTPATIENTS. Of the person who is
OUTPATIENTS at the beginning of the week, 85% will remain as OUTPATIENTS at the end
of the week, 10% will be admitted to WARDS, and 5% will be admitted to the ICU. Find the
expected cost in steady state.
0.50 0.50 0.00
Solution: The Transition Probability matrix is P = 0.10 0.50 0.40
0.05 0.10 0.85
Let S = p1 p2 p3 with p1 + p2 + p3 = 1, be the stationary distribution such that SP = S
0.50 0.50 0.00
i.e., p1 p2 p3 0.10 0.50 0.40 = p1 p2 p3
0.05 0.10 0.85
implies 0.50 p1 + 0.10 p2 + 0.05 p3 = p1
−0.50 p1 + 0.10 p2 + 0.05 p3 = 0 → (1)
0.50 p1 + 0.50 p2 + 0.10 p3 = p2
0.50 p1 − 0.50 p2 + 0.10 p3 = 0 → (2)
0.00 p1 + 0.40 p2 + 0.85 p3 = p3
DSSM Compiled by: Magisetty Harikrushna 43
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
0.00 p1 + 0.10 p2 − 0.15 p3 = 0 → (3)
p1 + p2 + p3 = 1 → (4)
By solving (1), (2) and (4), we get p1 = 0.11290, p2 = 0.24194, p3 = 0.64516
Now the cost of the policy is 0.11290 1000 + 0.24194 500
44 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A manufacturing company has a certain piece of equipment that is inspected at the
end of each day and classified as just overhauled, good, fair or inoperative. If the item is
inoperative it is overhauled, a procedure that takes one day. Let us denote the four
classifications as states 1, 2, 3 and 4 respectively. Assume that the working condition of the
equipment follows a Markov process with the following transition matrix:
Tomorrow
1 2 3 4
1 0 34 14 0
1 1
Today 2 0 2 2 0
3 0 0 12 12
4
1 0 0 0
If it costs Rs. 1,250 to overhaul a machine (including lost time), on the average, and Rs. 750 in
production is lost if a machine is found inoperative. Using steady state probabilities, compute
the expected per day cost of maintenance.
Solution: Here the given matrix P is an ergodic regular Markov process, it will certainly reach
to steady state equilibrium.
Let the steady state probabilities p1 , p2 , p3 and p4 represent the proportion of times that the
machine will be in states 1, 2, 3, and 4 respectively.
Now, with the steady state equations SP = S , here S = p1 p2 p3 p4
0 34 14 0
0 12 12 0
i.e., p1 p2 p3 p4 = p1 p2 p3 p4
0 0 12 12
1 0 0 0
implies p1 = p4 → (1)
3 1
p1 + p2 = p2
4 2
3 1
p1 = p2
4 2
3 3
p2 = p1 p2 = p4 → (2)
2 2
DSSM Compiled by: Magisetty Harikrushna 45
Reference Material MAPTSMS; MIAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.)
1 1 1
p1 + p2 + p3 = p3
4 2 2
1 1 1
p1 + p2 − p3 = 0 → (3)
4 2 2
1
p3 = p4
2
p3 = 2 p4 → (4) and
p1 + p2 + p3 + p4 = 1 → (5)
From (1), (2), (4) and (5)
3
p4 + p4 + 2 p4 + p4 = 1
2
11
p4 = 1
2
2
p4 =
11
2 3 4
p1 = ; p2 = ; p3 =
11 11 11
Thus, on an average, 2 out of every 11 days the machine will be overhauled; 3 out of every
11 days it will be in good condition; 4 out of every 11 days it will be in fair condition; and 2
out of every 11 days it will be found inoperative at the end of the days.
Hence the average cost per day of maintenance will be p1 1250 + p4 750
2 2
i.e., 1250 + 750 = Rs. 363.6
11 11
46 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; IAENG; MAMS M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Deterministic & Stochastic Statistical Methods
Unit – IV
Multivariate Distribution Theory
Univariate Normal Distribution
The probability density function of the Univariate normal distribution ( p = 1 variables) is given
1 x−
2
1 −
by f ( x ) = e 2 for − x
2
Here = E ( x ) = Mean and 2 = Var ( x ) = Variance, the parameters that completely
characterize the distribution.
Bivariate Normal Distribution
The probability density function of the Bivariate normal distribution
( p = 2 variables) is given by
−
1
2
x − x − 2 ( )
( x−x ) y − y y − y
+
2
1 (
2 1− 2 ) x x y y
f xy ( x, y ) =
e
2 x y 1 − 2
Y
Where x & y , x , y 0 and ( −1,1)
Two random variables X and Y are said to have the standard bivariate normal distribution withX
correlation coefficient ( −1,1) if their joint probability density function is given by
1 x 2 − 2 xy + y 2
−
f xy ( x, y ) =
1 ( )
2 1− 2
e where x = y = 0, x = y = 1
2 1 − 2
DSSM Compiled by: Magisetty Harikrushna 1
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Generalization to Multivariate Normal
The exponent term, squared statistical distance between x & in standard deviation units is
x−
2
= ( x − ) ( x − )
−1
Generalization to p 1 variables, here X p1 and parameters μ p1 and Σ p p then the exponent
term for multivariate normal is ( X − μ ) Σ−1 ( X − μ ) for − x and i = 1, 2,... p
T
1
The integral ... exp − ( X − μ ) Σ −1 ( X − μ ) = ( 2 ) Σ
T p /2 1/2
x1 xp 2
This implies that the multivariate normal distribution probability density function is given by
1
− ( X −μ ) −1 ( X −μ )
T
1
f ( X) = e 2
where − xi and i = 1, 2,... p .
( 2 )
p /2 1/2
Σ
It is denoted by X N p ( μ, Σ )
x1 1 11 12 . . . 1 p
x2 2 21 22 . . . 2p
. . . . . . . .
Here X = , μ = and Σ =
. . . . . . . .
. . . . . . . .
. pp
xn n p1 p 2 . .
Properties of Multivariate Normal Distribution
01. Joint Density: The multivariate normal distribution M N ( μ, Σ ) has joint density
( y − ) −1 ( y − )
1
1 1 −
f y (Y / μ, Σ ) = e 2
( 2 )
n /2 1/2
Σ
02. The contours of the joint distributions are n-dimensional ellipsoids
03. The M N ( μ, Σ ) joint distribution is specified by Mean μ and covariance Σ only
04. Moment Generating Function: The M N ( μ, Σ ) has moment generating distribution that
T 1 T
is given by M ( t ) = e
t + t t
2
where t is n1 real vector ( )
M x ( t ) = E et
T
x
2 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
05. Characteristic function: The M N ( μ, Σ ) has characteristic function that is given by
T 1 T
i t − t t
(t ) = e 2
where t is n1 real vector
06. Linear Combination: Let X follow multivariate normal distribution i.e., X M N ( μ, Σ )
r12 − r13r23
then any linear transformation Y = AX + bR12.3 = also normally
1 − r132 1 − r232
distributed and is Y M N ( Aμ + b, AΣAT )
DSSM Compiled by: Magisetty Harikrushna 3
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Joint Distribution – Discrete Type of Random Variables
Marginal distribution and Conditional distribution
Marginal distribution
Let X and Y be two discrete types of random variables, with joint distribution function
P ( x, y ) then,
The marginal distribution of X is PX ( x ) = PXY ( x, y )
y
The marginal distribution of Y is PY ( y ) = PXY ( x, y )
x
Here PX ( x ) 0; PY y 0 and P ( x ) = 1; P ( y ) = 1
X
X
Y
Y
Conditional probability function
Let X and Y be two discrete types of random variables then,
PXY ( x, y )
The conditional probability mass function of X given Y = y is PX /Y ( x / y ) = or
PY ( y )
P ( X = x, Y = y )
PX /Y ( x / y ) =
P (Y = y )
PXY ( x, y )
The conditional probability mass function of Y given X = x is PY / X ( y / x ) = or
PX ( x )
P ( X = x, Y = y )
PY / X ( y / x ) =
P ( X = x)
Here PX /Y ( x / y ) 0; PY / X ( y / x ) 0 and
P ( x, y )
XY
P ( y) P ( x, y )
XY
PX ( x )
PX /Y ( x / y ) =
X
X
PY ( y )
= Y
PY ( y )
= 1; PY / X ( y / x ) =
Y
Y
PX ( x )
=
PX ( x )
=1
4 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Note: If X and Y are independent variables then
PXY ( x, y ) = PX ( x ) PY ( y )
PXY ( x, y ) PX ( x ) PY ( y )
PX /Y ( x / y ) = = = PX ( x )
PY ( y ) PY ( y )
PXY ( x, y ) PX ( x ) PY ( y )
PY / X ( y / x ) = = = PY ( y )
PX ( x ) PX ( x )
E ( aX + bY ) = aE ( X ) + bE (Y )
V ( aX + bY ) = a 2V ( X ) + b2V (Y )
DSSM Compiled by: Magisetty Harikrushna 5
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Problem: For the joint probability distribution of two random variables X and Y given below
Y
1 2 3 4 Total
X
4 3 2 1 10
1
36 36 36 36 36
1 3 3 2 9
2
36 36 36 36 36
5 1 1 1 8
3
36 36 36 36 36
1 2 1 5 9
4
36 36 36 36 36
11 9 7 9
Total 1
36 36 36 36
Find (i) Marginal distribution of X and Y.
(ii) Conditional distribution of X given Y = 1 and Conditional distribution of Y given
X = 2.
Solution: Marginal distribution of X is P ( X = x ) = P ( X = x, Y = yi )
y
i.e., P ( X = 1) = P ( X = 1, Y = 1) + P ( X = 1, Y = 2 ) + P ( X = 1, Y = 3) + P ( X = 1, Y = 4 )
4 3 2 1 10
P ( X = 1) = + + + =
36 36 36 36 36
P ( X = 2 ) = P ( X = 2, Y = 1) + P ( X = 2, Y = 2 ) + P ( X = 2, Y = 3) + P ( X = 2, Y = 4 )
1 3 3 2 9
P ( X = 2) = + + + =
36 36 36 36 36
P ( X = 3) = P ( X = 3, Y = 1) + P ( X = 3, Y = 2 ) + P ( X = 3, Y = 3) + P ( X = 3, Y = 4 )
5 1 1 1 8
P ( X = 3) = + + + =
36 36 36 36 36
6 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
P ( X = 4 ) = P ( X = 4, Y = 1) + P ( X = 4, Y = 2 ) + P ( X = 4, Y = 3) + P ( X = 4, Y = 4 )
1 2 1 5 9
P ( X = 4) = + + + =
36 36 36 36 36
Therefore, Marginal distribution of X
X 1 2 3 4
10 9 8 9
P ( X)
36 36 36 36
Marginal distribution of Y is P (Y = y ) = P ( X = xi , Y = y )
x
i.e., P (Y = 1) = P ( X = 1, Y = 1) + P ( X = 2, Y = 1) + P ( X = 3, Y = 1) + P ( X = 4, Y = 1)
4 1 5 1 11
P (Y = 1) = + + + =
36 36 36 36 36
P (Y = 2 ) = P ( X = 1, Y = 2 ) + P ( X = 2, Y = 2 ) + P ( X = 3, Y = 2 ) + P ( X = 4, Y = 2 )
3 3 1 2 9
P (Y = 2 ) = + + + =
36 36 36 36 36
P (Y = 3) = P ( X = 1, Y = 3) + P ( X = 2, Y = 3) + P ( X = 3, Y = 3) + P ( X = 4, Y = 3)
2 3 1 1 7
P (Y = 3) = + + + =
36 36 36 36 36
P (Y = 4 ) = P ( X = 1, Y = 4 ) + P ( X = 2, Y = 4 ) + P ( X = 3, Y = 4 ) + P ( X = 4, Y = 4 )
1 2 1 5 9
P (Y = 4 ) = + + + =
36 36 36 36 36
Therefore, Marginal distribution of Y
Y 1 2 3 4
11 9 7 9
P (Y )
36 36 36 36
DSSM Compiled by: Magisetty Harikrushna 7
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
P ( X = xi , Y = y )
(ii) Conditional distribution of X given Y = y is P ( X = xi / Y = y ) =
P (Y = y )
4
P ( X = 1, Y = 1) 36 4
i.e., P ( X = 1/ Y = 1) = = =
P (Y = 1) 11 11
36
1
P ( X = 2, Y = 1) 36 1
P ( X = 2 / Y = 1) = = =
P (Y = 1) 11 11
36
5
P ( X = 3, Y = 1) 36 5
P ( X = 3 / Y = 1) = = =
P (Y = 1) 11 11
36
1
P ( X = 4, Y = 1) 36 1
P ( X = 4 / Y = 1) = = =
P (Y = 1) 11 11
36
Therefore, the Conditional distribution of X given Y = 1 is
X 1 2 3 4
4 1 5 1
P ( X / Y = 1)
11 11 11 11
P ( X = x, Y = yi )
Conditional distribution of Y given X = x is P (Y = yi / X = x ) =
P ( X = x)
1
P ( X = 2, Y = 1) 36 1
i.e., P (Y = 1/ X = 2 ) = = =
P ( X = 2) 9 9
36
3
P ( X = 2, Y = 2 ) 36 3
P (Y = 2 / X = 2 ) = = =
P ( X = 2) 9 9
36
8 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
3
P ( X = 2, Y = 3) 36 3
P (Y = 3 / X = 2 ) = = =
P ( X = 2) 9 9
36
2
P ( X = 2, Y = 4 ) 36 2
P (Y = 4 / X = 2 ) = = =
P ( X = 2) 9 9
36
Therefore, the Conditional distribution of Y given X = 2
Y 1 2 3 4
1 3 3 2
P ( Y / X = 2)
9 9 9 9
DSSM Compiled by: Magisetty Harikrushna 9
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Problem: Let X denote the number of speeding tickets issued by Officer on Monday, and let
Y denote the number given on Tuesday. The PMF of X and Y are,
X 0 1 2 3 4 Y 0 1 2 3 4 5
P(X) 0.1 0.2 0.3 0.2 0.2 P(Y) 0.1 0.1 0.2 0.3 0.2 0.1
Assume that X and Y are independent. Find the joint PMF of X and Y. Also find the PMF of
Z = X + Y.
Solution: Given X and Y are independent variables and the PMF of X and Y are,
X 0 1 2 3 4 Y 0 1 2 3 4 5
P(X) 0.1 0.2 0.3 0.2 0.2 P(Y) 0.1 0.1 0.2 0.3 0.2 0.1
So, the joint probability mass function is given by P ( X = x, Y = y ) = P ( X = x ) P (Y = y )
Since X and Y are independent variables.
P ( X = 0, Y = 0 ) = P ( X = 0 ) P (Y = 0 ) = 0.1 0.1 = 0.01
P ( X = 0, Y = 1) = P ( X = 0 ) P (Y = 1) = 0.1 0.1 = 0.01
P ( X = 0, Y = 2 ) = P ( X = 0 ) P (Y = 1) = 0.1 0.2 = 0.02
P ( X = 0, Y = 3) = P ( X = 0 ) P (Y = 3) = 0.1 0.3 = 0.03
P ( X = 0, Y = 4 ) = P ( X = 0 ) P (Y = 4 ) = 0.1 0.2 = 0.02
P ( X = 0, Y = 5) = P ( X = 0 ) P (Y = 5) = 0.1 0.1 = 0.01
10 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Therefore, the joint probability mass function of X and Y is
Y
0 1 2 3 4 5 Total
X
0 0.01 0.01 0.02 0.03 0.02 0.01 0.1
1 0.02 0.02 0.04 0.06 0.04 0.02 0.2
2 0.03 0.03 0.06 0.09 0.06 0.03 0.3
3 0.02 0.02 0.04 0.06 0.04 0.02 0.2
4 0.02 0.02 0.04 0.06 0.04 0.02 0.2
Total 0.1 0.1 0.2 0.3 0.2 0.1 1
Let Z = X + Y
The possible values of Z = X + Y are 0, 1, 2, 3, 4, 5, 6, 7, 8, 9.
P ( Z = 0 ) = P ( X = 0, Y = 0 ) = 0.01
P ( Z = 1) = P ( X = 0, Y = 1) or ( X = 1, Y = 0 )
P ( Z = 1) = P ( X = 0, Y = 1) + P ( X = 1, Y = 0 )
P ( Z = 1) = 0.01 + 0.02 = 0.03
P ( Z = 2 ) = P ( X = 0, Y = 2 ) or ( X = 2, Y = 0 ) or ( X = 1, Y = 1)
P ( Z = 2 ) = P ( X = 0, Y = 2 ) + P ( X = 2, Y = 0 ) + P ( X = 1, Y = 1)
P ( Z = 2 ) = 0.02 + 0.03 + 0.02 = 0.07
P ( Z = 3) = P ( X = 0, Y = 3) or ( X = 3, Y = 0 ) or ( X = 1, Y = 2 ) or ( X = 2, Y = 1)
P ( Z = 3) = P ( X = 0, Y = 3) + P ( X = 3, Y = 0 ) + P ( X = 1, Y = 2 ) + P ( X = 2, Y = 1)
P ( Z = 3) = 0.03 + 0.02 + 0.04 + 0.03 = 0.12
P ( Z = 4 ) = P ( X = 0, Y = 4 ) or ( X = 4, Y = 0 ) or ( X = 1, Y = 3) or ( X = 3, Y = 1) or ( X = 2, Y = 2 )
DSSM Compiled by: Magisetty Harikrushna 11
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
P ( Z = 4 ) = P ( X = 0, Y = 4 ) + P ( X = 4, Y = 0 ) + P ( X = 1, Y = 3) + P ( X = 3, Y = 1) + P ( X = 2, Y = 2 )
P ( Z = 4 ) = 0.02 + 0.02 + 0.06 + 0.02 + 0.06 = 0.18
Similarly,
P ( Z = 5) = 0 + 0.01 + 0.04 + 0.02 + 0.09 + 0.04 = 0.20
P ( Z = 6 ) = 0.02 + 0.00 + 0.06 + 0.04 + 0.06 = 0.18
P ( Z = 7 ) = 0.03 + 0.00 + 0.04 + 0.06 = 0.13
P ( Z = 8) = 0.02 + 0.00 + 0.04 = 0.06
P ( Z = 9 ) = 0.02 + 0.00 = 0.02
The probability mass function of Z is
Z 0 1 2 3 4 5 6 7 8 9
P(Z) 0.01 0.03 0.07 0.12 0.18 0.20 0.18 0.13 0.06 0.02
12 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Let X and Y be two independent random variables with the following probability
distributions
X P(X) Y P(Y)
1 0.3 0 0.4
2 0.5 1 0.6
3 0.2
(a) Find the probability mass function (PMF) of the random variable Z = X + Y .
(b) Calculate the expected value (mean) and variance of Z
Solution: Given X and Y be two independent random variables with the following
probability distributions
X P(X) Y P(Y)
1 0.3 0 0.4
2 0.5 1 0.6
3 0.2
So, the joint probability mass function is given by P ( X = x, Y = y ) = P ( X = x ) P (Y = y )
Since X and Y are independent variables.
P ( X = 1, Y = 0 ) = P ( X = 1) P (Y = 0 ) = 0.3 0.4 = 0.12
P ( X = 1, Y = 1) = P ( X = 1) P (Y = 1) = 0.3 0.6 = 0.18
P ( X = 2, Y = 0 ) = P ( X = 2 ) P (Y = 0 ) = 0.5 0.4 = 0.20
P ( X = 2, Y = 1) = P ( X = 2 ) P (Y = 1) = 0.5 0.6 = 0.30
P ( X = 3, Y = 0 ) = P ( X = 3) P (Y = 0 ) = 0.2 0.4 = 0.08
P ( X = 3, Y = 1) = P ( X = 3) P (Y = 1) = 0.2 0.6 = 0.12
DSSM Compiled by: Magisetty Harikrushna 13
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Therefore, the joint probability mass function of X and Y is
Y
0 1 Total
X
1 0.12 0.18 0.30
2 0.20 0.30 0.50
3 0.08 0.12 0.20
Total 0.40 0.60 1
a) Let Z = X + Y
The possible values of Z = X + Y are 1, 2, 3, 4.
P ( Z = 1) = P ( X = 1, Y = 0 ) = 0.12
P ( Z = 2 ) = P ( X = 1, Y = 1) or ( X = 2, Y = 0 )
P ( Z = 2 ) = P ( X = 1, Y = 1) + P ( X = 2, Y = 0 )
P ( Z = 2 ) = 0.18 + 0.20 = 0.38
P ( Z = 3) = P ( X = 2, Y = 1) or ( X = 3, Y = 0 )
P ( Z = 3) = P ( X = 2, Y = 1) + P ( X = 3, Y = 0 )
P ( Z = 3) = 0.30 + 0.08 = 0.38
P ( Z = 4 ) = P ( X = 3, Y = 1)
P ( Z = 4 ) = 0.12
The probability mass function of Z is
Z 1 2 3 4
P(Z) 0.12 0.38 0.38 0.12
14 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
4
b) Expected value of Z is E ( Z ) = = Z P ( Z )
z =1
E ( Z ) = (1)( 0.12 ) + ( 2 )( 0.38) + ( 3)( 0.38) + ( 4 )( 0.12 )
Expected value of Z is E ( Z ) = 2.5
E (Z ) = Z P(Z )
4
2 2
z =1
E ( Z 2 ) = (1) ( 0.12 ) + ( 2 ) ( 0.38 ) + ( 3) ( 0.38 ) + ( 4 ) ( 0.12 )
2 2 2 2
E ( Z 2 ) = 6.98
( )
Variance of Z is V ( Z ) = E Z 2 − ( E ( Z ) )
2
V ( Z ) = 6.98 − 6.25 = 0.73
Variance of Z is V ( Z ) = 0.73
DSSM Compiled by: Magisetty Harikrushna 15
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Continuous Type of Random Variables
Marginal distribution and Conditional distribution
Marginal distribution
Let X and Y be two continuous types of random variables, with joint distribution function
f ( x, y ) then,
The marginal distribution of X is f X ( x ) = f XY ( x, y ) dy
−
The marginal distribution of Y is fY ( y ) = f XY ( x, y ) dx
−
Here f X ( x ) 0; fY y 0 and
−
f ( x ) = 1; f ( y ) = 1
X
−
Y
Conditional probability function
Let X and Y be two continuous types of random variables then,
f XY ( x, y )
The conditional probability mass function of X given Y = y is f X /Y ( x / y ) =
fY ( y )
f XY ( x, y )
The conditional probability mass function of Y given X = x is fY / X ( y / x ) =
fX ( x)
Here f X /Y ( x / y ) 0; fY / X ( y / x ) 0 and f X /Y ( x / y ) dx = 1 ; fY / X ( y / x ) dy = 1
− −
Note: If X and Y are independent variables then
f XY ( x, y ) = f X ( x ) fY ( y )
f XY ( x, y ) f X ( x ) fY ( y )
f X /Y ( x / y ) = = = fX ( x)
fY ( y ) fY ( y )
f XY ( x, y ) f X ( x ) fY ( y )
fY / X ( y / x ) = = = fY ( y )
fX ( x) fX ( x)
16 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Let X and Y be two random variables with joint probability density function
kxy 0 x y 1
f XY ( x, y ) = find the value of ‘k’ and also find the marginal functions.
0 Other wise
kxy 0 x y 1
Solution: Given joint probability density function f XY ( x, y ) =
0 Other wise
We have f XY ( x, y ) dxdy = 1
− −
1 1
i.e., f XY ( x, y ) dxdy = 1
x =0 y = x
1 1
x=0 y= x kxy dy dx = 1
1
1
y2
2 dx = 1
x =0
kx
x
1
1 x2
2 − 2 dx = 1
x =0
kx
1
k
2 x =0
( x − x3 ) dx = 1
1
k x2 x4
− =1
2 2 4 0
k 1 1
− =1
22 4
k 1
=1
24
k =8
8xy 0 x y 1
f XY ( x, y ) =
0 Other wise
DSSM Compiled by: Magisetty Harikrushna 17
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
The marginal distribution of X is f X ( x ) = f XY ( x, y ) dy
−
1
fX ( x) = 8xy dy
y=x
1
y2
f X ( x ) = 8x
2 x
1 x2
f X ( x ) = 8x −
2 2
4 x (1 − x 2 ) 0 x 1
The marginal distribution of X is f X ( x ) =
0 Other wise
The marginal distribution of Y is fY ( y ) = f XY ( x, y ) dx
−
fY ( y ) = 8xy dx
x =0
y
x2
fY ( y ) = 8 y
2 0
y2
fY ( y ) = 8 y − 0
2
fY ( y ) = 4 y 3
4 y 3 0 y 1
The marginal distribution of Y is fY ( y ) =
0 Other wise
18 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Let X and Y be two random variables with joint probability density function
k ( x y + xy ) 0 x 1, 0 y 1
2 3
f XY ( x, y ) = find the value of ‘k’ and also find the marginal
0 Other wise
functions.
Problem: Let X and Y be two random variables with joint probability density function
k ( x y + xy ) x + y 1, x 0, y 0
2 3
f XY ( x, y ) = find the value of ‘k’ and also find the
0 Other wise
marginal functions.
DSSM Compiled by: Magisetty Harikrushna 19
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Problem: Let X, Y and Z denote 3 jointly distributed random variable with joint density
K x + yz
function f ( x, y, z ) =
2
( ) 0 x 1, 0 y 1, 0 z 1
then
0 otherwise
(i) Find the value of K.
(ii) Determine the marginal distributions of X, Y and Z.
(iii) Determine the joint marginal distributions of X , Y ; X , Z ; Y , Z .
(iv) The conditional distribution of Z given X = x, Y = y
(v) Determine the conditional expectation of U = X 2 + Y + Z given X = x, Y = y .
(vi) Determine E ( XYZ )
K x + yz
Solution: Given joint density function f ( x, y, z ) =
2
( ) 0 x 1, 0 y 1, 0 z 1
0 otherwise
(i) The Value of ‘K’
We have f ( x, y, z ) dx dy dz = 1
− − −
1 1 1
i.e., f ( x, y, z ) dx dy dz = 1
x =0 y =0 z =0
1 1 1
K (x )
+ yz dx dy dz = 1
2
x =0 y =0 z =0
1 1
1 2
K x=0 y=0 z=0 ( x + yz ) dz dx dy = 1
1
1
21
z2
K x z + y dx dy = 1
x =0 y =0
2 0
1
1 1
x + y dx dy = 1
2
K
x =0 y =0
2
1 1 2 1
K x + 2 y dy dx = 1
x =0 y =0
20 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1
1
1 y2
K x2 y + dx = 1
x =0 0
2 2
1
1
x + dx = 1
2
K
x =0
4
1
x3 1
K + x =1
3 4 0
1 1
K + =1
3 4
7
K =1
12
12
K =
7
12 2
f ( x, y , z ) = 7
x + yz ( ) 0 x 1, 0 y 1, 0 z 1
0 otherwise
(ii) Marginal distribution of X is f X ( x ) = f ( x, y, z ) dy dz
y =− z =−
1 1
fX ( x) = f ( x, y, z ) dy dz
y =0 z =0
1 1
f X ( x) =
12 2
7
( )
x + yz dy dz
y =0 z =0
12
1 1
fX ( x) = ( x + yz ) dz dy
2
7 y =0 z =0
1
12 2 z2
1
fX ( x) =
7 y=0
x z + y dy
2 0
12 2 1
1
fX ( x) = x + y dy
7 y =0 2
1
12 y2 1
f X ( x ) = x2 y +
7 2 2 0
DSSM Compiled by: Magisetty Harikrushna 21
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
12 2 1 1
f X ( x) = x +
7 2 2
12 2 1
fX ( x) = x + for 0 x 1
7 4
Marginal distribution of Y is fY ( y ) = f ( x, y, z ) dx dz
x =− z =−
1 1
fY ( y ) = f ( x, y, z ) dx dz
x =0 z =0
1 1
fY ( y ) = 7
( x + yz ) dx dz
12 2
x =0 z =0
12
1 1
fY ( y ) = ( x + yz ) dz dx
2
7 x =0 z =0
1
12 2 z2
1
fY ( y ) =
7 x=0
x z + y dx
2 0
12 2 1
1
fY ( y ) = x + y dx
7 x =0 2
1
12 x3 y
fY ( y ) = + x
7 3 2 0
12 1 y
fY ( y ) = + for 0 y 1
7 3 2
Marginal distribution of Z Y is f Z ( z ) = f ( x, y, z ) dx dy
x =− y =−
1 1
fZ ( z ) = f ( x, y, z ) dx dy
x =0 y =0
1 1
fZ ( z ) = 7
(
12 2
x + yz dx dy)
x =0 y =0
12
1 1
fZ ( z ) =
7 x=0 y=0
x (
2
+ yz dy )
dx
1
12 2 y2
1
fZ ( z ) =
7 x=0
x y + z dx
2 0
22 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
12 2 1
1
fZ ( z ) = x + z dx
7 x =0 2
1
12 x3 1
fZ ( z ) = + z x
7 3 2 0
12 1 z
fZ ( z ) = + for 0 z 1
7 3 2
(iii) The joint marginal distributions of X , Y is f XY ( x, y ) = f ( x, y, z ) dz
z =−
1
f XY ( x, y ) = 7
( x + yz ) dz
12 2
z =0
1
12 z2
f XY ( x, y ) = x 2 z + y
7 2 0
12 2 1
f XY ( x, y ) = x + y
7 2
12 2 y
f XY ( x, y ) = x + for 0 x 1; 0 y 1
7 2
The joint marginal distributions of X , Z is f XZ ( x, z ) = f ( x, y, z ) dy
y =−
1
f XZ ( x, z ) = 7
( x + yz ) dy
12 2
y =0
1
12 y2
f XZ ( x, z ) = x 2 y + z
7 2 0
12 2 1
f XZ ( x, z ) = x +z
7 2
12 2 z
f XZ ( x, z ) = x + for 0 x 1; 0 z 1
7 2
The joint marginal distributions of Y , Z is fYZ ( y, z ) = f ( x, y, z ) dx
x =−
1
fYZ ( y, z ) = 7
( x + yz ) dx
12 2
x =0
1
12 x3
fYZ ( y, z ) = + yzx
7 3 0
DSSM Compiled by: Magisetty Harikrushna 23
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
12 1
fYZ ( y, z ) = + yz for 0 y 1; 0 z 1
7 3
f ( x, y , z )
(iv) The conditional distribution of Z given X = x, Y = y is
f XY ( x, y )
f ( x, y , z )
12 2
7
( x + yz )
=
f XY ( x, y ) 12 2 1
x + y
7 2
f ( x, y , z ) x 2 + yz
= for 0 z 1
f XY ( x, y ) x + y
2 1
2
(v) Conditional expectation of U = X 2 +Y + Z given X = x, Y = y is given by
f ( x, y , z )
E U x, y =
z =−
U
f XY ( x, y )
dz
+ yz
1 2
E U x, y = (x + y+z )x
2
dz
1
z =0 x + y
2
2
1
(x )( )
1
= 2
+ y + z x 2 + yz dz
1
x + y z =0
2
( yz ))
1
=
1
1
2
( ) (
+ y x 2 + y + x 2 z + x 2 x 2 + y dz
x2 + y 0
2
1
z3 2
=
1
1
y +
y x 2
+ y + x 2 z
2 + x 2
(
x 2
+ y z ) ( )
x2 + y 3 0
2
1
=
1
1
(
y + y x 2 + y + x 2 + x 2 x 2 + y ) ( )
x2 + y 3
1 2
2
y x2 2 1 2
Conditional expectation of U given X = x, Y = y is
1
+ + x + y x + y ( )
x2 + y 3 2 2
1
2
(vi) E ( XYZ ) = xyz f ( x, y, z ) dx dy dz
− − −
24 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
1 1 1
xyz ( x 2 + yz ) dx dy dz
12
E ( XYZ ) =
7 x =0 y =0 z =0
12
1 1
1 3
E ( XYZ ) = x yz + xy z dz dx dy
2 2
7 x =0 y =0 z =0
1
12
1 1
3 z2 2 z
3
E ( XYZ ) = x y + xy dx dy
7 x =0 y =0 2 3 0
3 1 2 1
1 1
12
E ( XYZ ) = x y + xy dx dy
7 x =0 y =0 2 3
12 3 1 1
1 1
E ( XYZ ) = x y + xy 2 dy dx
7 x =0 y =0 2 3
1
12 3 y 2 1 y3 1
1
E ( XYZ ) =
7 x=0
x + x dx
2 2 3 3 0
12 3 1 1 1 1
1
E ( XYZ ) = x +x dx
7 x =0 2 2 3 3
12 3 1 1
1
E ( XYZ ) = x + x dx
7 x =0 4 9
1
12 x 4 1 x 2 1
E ( XYZ ) = +
7 4 4 2 9 0
12 1 1
E ( XYZ ) = +
7 16 18
17
E ( XYZ ) =
84
DSSM Compiled by: Magisetty Harikrushna 25
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Problem: Suppose that a rectangle is constructed by first choosing its length, X and then
choosing its width Y. Its length X is selected form an exponential distribution with mean
1
= = 5 . Once the length has been chosen its width, Y, is selected from a uniform distribution
form 0 to half its length.
(i) Find the mean and variance of the area of the rectangle A = XY
(ii) Find the probability that its area A = XY is less than 4.
Solution: Given the length X and width Y of the rectangle.
1
Length X is selected form an exponential distribution with mean = = 5,
1 −1 x 1 1
− x
So, the marginal distribution of X is f X ( x ) = e 5 for x 0 f ( x) = e
,x0
5
Once the length has been chosen, its width Y, is selected from a uniform distribution from 0 to
half its length.
1
So, the conditional distribution of Y given X = x is fY / X ( y / x ) = for 0 y x / 2
x/2
f XY ( x, y )
The joint density function is f XY ( x, y ) = fY ( y ) f X /Y ( x / y ) f X /Y ( x / y ) =
fY ( y )
2 1 − 15 x
f XY ( x, y ) = e
x 5
2 − 15 x
f XY ( x, y ) = e for 0 y x / 2; x 0
5x
The area is A = XY
(i) The mean of the area of the rectangle is E ( A) = E ( XY )
E ( A) = E ( XY ) = xy f ( x, y ) dxdy
XY
− −
2 − 15 x
E ( A) = xy e dxdy
− −
5x
26 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
2
x /2 1
− x
E ( A) = y e 5 dy dx
5 x =0 y =0
2
x /2 1
− x
E ( A) = y e 5
dy dx
5 x =0 y =0
x /2
− x y
1 2
2
E ( A) = e 5 dx
5 x =0 2 0
− x x
1 2
1
E ( A ) = e dx
5
5 x =0 4
1
1 − x
E ( A) =
20 x=0
e 5
x3−1 dx
1 ( 3) Γ n
E ( A) = e kx
x n 1dx
20 1 3 0 kn
5
2! 53 25
E ( A) = = ( n + 1) = n !
20 2
E(A ) = E(X Y
2 2 2
)= x y 2 2
f XY ( x, y ) dxdy
− −
2 − 15 x
E(A ) = x y
2 2
e dxdy 2
− −
5x
2
x /2 1
E ( A ) = xy e 5 dy dx
− x
2 2
5 x =0 y =0
− x
1 x /2
E ( A ) = xe y 2 dy dx
2 2 5
5 x =0
y =0
x /2
− x y
1 3
E ( A ) = xe 5 dx
2 2
5 x =0 3 0
− x x
1 3
E(A ) =
2
15 x=0
2
dx
5
xe
8
DSSM Compiled by: Magisetty Harikrushna 27
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
1
E(A ) =
1 − x
60 x=0
2 4
x e 5 dx
1
E ( A2 ) =
1 − x
60 x=0
e 5
x5−1 dx
1 ( 5) Γ n
E ( A2 ) = e kx
x n 1dx
60 1 5 0 kn
5
4! 55
E ( A2 ) = = 1250 ( n + 1) = n !
60
(ii) The probability that its area A = XY is less than 4 is form the diagram we have,
2 2 x /2 4/ x
P ( A 4 ) = P ( XY 4 ) = f ( xy ) dxdy + f ( xy ) dxdy
x =0 y =0 x = 2 2 y =0
2 2 x /2 4/ x
2 − 15 x 2 − 15 x
P ( A 4) = e dxdy + 5x e dxdy
x =0 y =0
5x x=2 2 y =0
2 1 − 15 x 1 − 15 x
2 2 x /2 4/ x
2
=
5 x =0 x
e
1 dy
dx +
5 x =2 2 x
e
1 dy dx
y =0 y =0
2 2
2 1 − 15 x 2 1 − 15 x
= e ( y )0 dx + e ( y )0 dx
x /2 4/ x
5 x =0 x 5 x=2 2 x
2 1 − 15 x x 1 − 15 x 4
2 2
2
5 x=0 x 5 x =2 2 x
= e dx + e dx
2 x
2 2 −1 −1
1 x 8 x
−2 5
= e 5
dx + x e dx
5 x =0 5 x=2 2
2 2 2 2
−1 x −1,
1e5 8 5
= + By Numerical Method
5 −1 5 5
5 0
28 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
−2 2 8
= − e 5 − e0 + ( −0.10335)
5
P ( A 4 ) = P ( XY 4 ) = 0.2184
DSSM Compiled by: Magisetty Harikrushna 29
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Coefficient of Multiple Correlation X 1 X 2 X 3
The coefficient of multiple correlation, denoted R, is a scalar that is defined as the
Pearson correlation coefficient between the predicted and the actual values of the dependent
variable in a linear regression model that includes an intercept.
Suppose, X 1 , X 2 and X 3 are three variables having observations on N entities. Then
multiple correlation coefficient of X 1 on X 2 and X 3 is the simple correlation coefficient
between X 1 and the joint effect of X 2 and X 3 . It can also be defined as the correlation
between X 1 and its estimate based on X 2 and X 3 .
Multiple correlation coefficient is the simple correlation coefficient between a variable and its
estimate and is given by
r122 + r132 − 2r12 r13r23 r122 + r232 − 2r12 r13r23 r132 + r232 − 2r12 r13r23
R1.23 = ; R2.13 = and R3.12 =
1 − r232 1 − r132 1 − r122
N ( X 1 X 2 ) − ( X 1 )( X 2 )
Here r12 = …
N ( X 12 ) − ( X 1 ) N ( X 22 ) − ( X 2 )
2 2
Coefficient of Partial Correlation
Let us consider the case of three variables X 1 X 2 and X 3 . Sometimes the correlation between
two variables X 1 and X 2 may be partly due to the correlation of a third variable X 3 with both
X 1 and X 2 . In this point of situation, one may be absorbed to study the correlation between
X 1 and X 2 when the effect of X 3 on each of X 1 and X 2 is eliminated. This correlation is
known as partial correlation. The correlation coefficient between X 1 and X 2 after eliminating
the linear effect of X 3 on X 1 and X 2 is called the partial correlation coefficient.
30 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
The partial correlation coefficient between X 1 and X 2 when the effect of X 3 on each of X 1
r12 − r13r23
and X 2 are eliminated is R12.3 =
1 − r132 1 − r232
The partial correlation coefficient between X 1 and X 3 , when the effect of X 2 on each of X 1
r13 − r12 r23
and X 3 are eliminated is R13.2 =
1 − r122 1 − r232
The partial correlation coefficient between X 2 and X 3 , when the effect of X 1 on each of X 2
r23 − r12 r13
and X 3 are eliminated is R23.1 =
1 − r122 1 − r132
DSSM Compiled by: Magisetty Harikrushna 31
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Problem: Given data X 1 = 2, 5, 7, 8, 5; X 2 = 8, 8, 6, 5, 3; X 3 = 0, 1, 1, 3, 4. Calculate the partial
correlations co-efficient.
Solution: Given data X 1 = 2, 5, 7, 8, 5; X 2 = 8, 8, 6, 5, 3; X 3 = 0, 1, 1, 3, 4.
X1 X2 X3 X1 X 2 X1 X 3 X2 X3 X12 X 22 X 32
2 8 0 16 0 0 4 64 0
5 8 1 40 5 8 25 64 1
7 6 1 42 7 6 49 36 1
8 5 3 40 24 15 64 25 9
5 3 4 15 20 12 25 9 16
= 27 = 30 = 9 = 153 = 56 = 41 = 167 = 198 = 27
The zero order correlation coefficients are
N ( X 1 X 2 ) − ( X 1 )( X 2 )
r12 =
N ( X 12 ) − ( X 1 ) N ( X 22 ) − ( X 2 )
2 2
( 5) (153) − ( 27 )( 30 )
r12 = = −0.4607
5 (167 ) − ( 27 ) 5 (198 ) − ( 30 )
2 2
N ( X 1 X 3 ) − ( X 1 )( X 3 )
r13 =
N ( X 12 ) − ( X 1 ) N ( X 32 ) − ( X 3 )
2 2
5 ( 56 ) − ( 27 )( 9 )
r13 = = 0.4890
5 (167 ) − ( 27 ) 5 ( 27 ) − ( 9 )
2 2
N ( X 2 X 3 ) − ( X 2 )( X 3 )
r23 =
N ( X 22 ) − ( X 2 ) N ( X 32 ) − ( X 3 )
2 2
5 ( 41) − ( 30 )( 9 )
r23 = = −0.9324
5 (198) − ( 30 ) 5 ( 27 ) − ( 9 )
2 2
The partial correlation coefficient between X 1 and X 2 when the effect of X 3 on each of X 1
r12 − r13r23
and X 2 are eliminated is R12.3 =
1 − r132 1 − r232
( −0.4607 ) − ( 0.4890 )( −0.9324 )
R12.3 = = −0.0150
1 − ( 0.4890 ) 1 − ( −0.9324 )
2 2
32 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
The partial correlation coefficient between X 1 and X 3 , when the effect of X 2 on each of X 1
r13 − r12 r23
and X 3 are eliminated is R13.2 =
1 − r122 1 − r232
( 0.4890 ) − ( −0.4607 )( −0.9324 )
R13.2 = = 0.1854
1 − ( −0.4607 ) 1 − ( −0.9324 )
2 2
The partial correlation coefficient between X 2 and X 3 , when the effect of X 1 on each of X 2
r23 − r12 r13
and X 3 are eliminated is R23.1 =
1 − r122 1 − r132
( −0.9324 ) − ( −0.4607 )( 0.4890 )
R23.1 = = −0.9133
1 − ( −0.4607 ) 1 − ( 0.4890 )
2 2
DSSM Compiled by: Magisetty Harikrushna 33
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Bit
Bit is a unit of information. 1 bit refers to the amount of information that one is uncertain
about in a binary random variable that takes the value of either 0 or 1 with equal probability.
Surprisal
Surprisal quantifies the uncertainty in a random variable X taking a certain value x based on
its probability of occurrence P ( X = x ) or P ( x ) . Surprisal is measured in bits when the base
of the logarithm is 2.
1
Base ' 2 '− Bit ;
i.e., s ( x ) = log Base ' e '− Nat ;
P( x )
2 = − log 2P( x )
Base '10 '− Hartly.
Entropy
Entropy of a random variable is denoted by H ( X ) is the expected or average surprisal based
on its probability distribution. Entropy is measured in bits when the base of logarithm is 2.
H ( X ) = P ( x) s ( x)
x
i.e., H ( X ) = − P ( x ) log 2
( p x)
Joint Entropy
The joint entropy of a pair of discrete random variables X , Y ~ P ( x, y ) is the amount of
information needed on average to specify both their values.
H ( X ) = − P ( x, y ) log 2p( x, y )
x y
34 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Conditional Entropy
The conditional entropy of a discrete random variable Y given X, for X , Y ~ P ( x, y ) ,
expresses how much extra information you still need to supply on average to communicate Y
given that the other party knows X.
H (Y / X ) = − P ( x, y ) log 2p( y / x )
x y
Similarly, H ( X / Y ) = − P ( x, y ) log 2p( x / y )
x y
Coding: The procedure of associating words from the alphabet of one language to the
alphabet of with the given words of another language.
i.e., we represent the words form one language into with the words of another language.
The objective of the coding is to increase the efficiency and reduce the transmission errors.
Let the source or the transmitter be represented by the set of containing ‘k’ symbols.
i.e., S = s1 , s2 , s3 , ... sk is known as source alphabet.
Let us consider another set X containing ‘r’ symbols, X = x1 , x2 , x3 , ... xr is called Code
alphabet.
Now each and every symbol of S is mapped into sequence of symbol of X.
For example:
s1 = x1 x2
s2 = x2 x1 x3
s3 = x3 x2 x4 x5
...
Discrete Memoryless Source (DMS)
A source from which the data is being emitted at successive intervals, which is independent
of previous values, can be characterized as discrete memoryless source.
DSSM Compiled by: Magisetty Harikrushna 35
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Source Coding Theorem
Let a DMS outputs a symbol every ‘t’ second, each symbol is selected from a finite
set of symbols xi , i = 1, 2,...L , occurring with probabilities P ( xi ) , i = 1, 2,...L . Then the
L
entropy of this DMS in bits per source symbol is H ( X ) = − P ( xi ) log 2 i log 2
( ) L p x
i =1
Fixed Length Code
If a finite set of symbols xi , i = 1, 2,...L , then the length of the codes is given by
log 2 + 1 If L is not power of 2
L
R=
L
log 2 f L is power of 2
Prefix Codes
A prefix code is one in which no codeword forms the prefix of any other codeword,
such codes are called as Prefix codes or instantaneous codes.
The rate of inference about the outcomes per second is given by Rate = L H ( X )
H (X )
The percentage of maximum possible information is given by
R
36 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Kullback–Leibler Divergence
It is a statistical measurement from information theory that is commonly used to
quantify the difference between one probability distribution from a reference probability
distribution.
KL divergence is a non-symmetric metric that measures the relative entropy or difference in
information represented by two distributions. It can be thought of as measuring the distance
between two data distributions showing how different the two distributions are from each
other.
➢ It is a measure of how one probability distribution is different from the second
➢ It is also called as relative entropy.
➢ Divergence is not distance.
For discrete type of random variables KL divergence is given by
P ( x)
The relative entropy from Q to P is DKL ( P ( X ) || Q ( X ) ) = P ( X ) log and
xX Q( X )
Q ( x)
The relative entropy from P to Q is DKL ( Q ( X ) || P ( X ) ) = Q ( X ) log
xX P( X )
For continuous type of random variables KL divergence is given by
P ( x)
The relative entropy from Q to P is DKL ( P ( X ) || Q ( X ) ) = P ( X ) log Q ( X ) dx and
−
Q ( x)
The relative entropy from P to Q is DKL ( Q ( X ) || P ( X ) ) = Q ( X ) log P ( X ) dx
−
DSSM Compiled by: Magisetty Harikrushna 37
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT (
M.Sc., B.Ed., M.Phil., Ph.D.)
Problem: Compute the KL divergence for the following
X 0 1 2
Distribution P(X) 0.36 0.48 0.16
Distribution Q(X) 0.333 0.333 0.333
P ( x)
Solution: The relative entropy from Q to P is DKL ( P ( X ) || Q ( X ) ) = P ( X ) log
xX Q( X )
2
P ( x)
DKL ( P ( X ) || Q ( X ) ) = P ( X ) log
x =0 Q( X )
0.36 0.48 0.16
= 0.36 log + 0.48 log + 0.16 log
0.33 0.33 0.33
Q ( x)
The relative entropy from P to Q is DKL ( Q ( X ) || P ( X ) ) = Q ( X ) log
xX P( X )
0.33 0.33 0.33
= 0.33 log + 0.33 log + 0.33 log
0.36 0.48 0.16
Problem: Compute the relative entropy from P to Q and from Q to P.
X 0 1 2
Distribution P(X) 11/25 8/25 6/25
Distribution Q(X) 1/3 1/3 1/3
38 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Deterministic & Stochastic Statistical Methods
Unit – V
Optimization
Hessian Matrix: The Hessian matrix, is a square matrix of second-order partial derivatives of
a scalar-valued function, it is denoted by H .
If f ( X ) is a scalar-valued function, where X = ( x1 , x2 ,...xn ) then the Hessian matrix is defined
2 f 2 f 2 f
x 2 x1x2
...
x1xn
1
2 f 2 f 2 f
as H = x2 x1 x22 x2 xn
2 f 2 f 2 f
x x xn x2 xn2
n 1
Here cross partials are equal for f ( X ) , so the matrix is symmetric.
The Hessian matrix helps determine the local behaviour of a function around critical points
such as maxima, minima, or saddle points. Specifically, it's used in optimization algorithms to
test for optimality conditions.
Principal Minor: Let H be n n matrix. A principal minor of H , of order k is the determinant
of the matrix formed by deleting some ( n − k ) rows and their corresponding columns of H ,
where k = 0,1, 2,...n − 1 .
Leading Principal Minor: The k th order leading principal minor (LPM) of H is the
determinant of the matrix formed by deleting the last ( n − k ) columns and rows of H .
Conditions for a Minimum or a Maximum Value of a Function of Several Variables
For the given function f ( X ) of several independent variables x
e Calculate f ( X) and set it to zero.
e Solve the equation set to get a solution vector x * .
e Calculate f ( X ) , evaluate it at x * .
2
DSSM Compiled by: Magisetty Harikrushna 1
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
e Inspect the Hessian matrix H ( X ) = 2
f ( X ) at point x * .
The matrix H is positive definite if all its leading principal minors are positive at x *
i.e., Leading principal minors H1 0; H 2 0; H 3 0... at x * then the function f ( X )
is called strictly convex, and the point x * is strict local minimum of the function f ( X ) .
The matrix H is negative definite if all its leading principal minors are not zero and
alternate in sign, but first leading principal minor should be negative.
i.e., H1 0; H 2 0; H 3 0... , at x * then the function f ( X ) is called strictly concave
and the point x * is strict local maximum of the function f ( X ) .
The matrix H is positive semi definite if all its leading principal minors are positive at x *
and at least one of H i = 0 for i = 2,3...n .
i.e., Leading principal minors H1 0; H 2 0; H 3 0... then the function f ( X ) is called
convex, and the point x * is global minimum of the function f ( X ) .
The matrix H is negative semi definite if all its leading principal minors alternate in sign,
but first leading principal minor should be negative and at least one of H i = 0 for
i = 2,3...n .
i.e., Leading principal minors H1 0; H 2 0; H 3 0... , etc. at x * then the function
f ( X ) is concave and the point x * is global maximum of the function f ( X )
Indefinite if none of the above cases happen.
Note:
A function which is both convex and concave, then it has to be linear function.
If f and g are two convex functions over the convex set then
f + g , f , ( 0 ) and max ( f , g ) are also convex functions.
2 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
If f and g are two concave functions over the convex set then
f + g , f , ( 0 ) and min ( f , g ) are also concave functions.
Problem: Find the maximum, minimum of the function f ( x ) = x3 − 6 x 2 + 9 x
Solution: The given function is f ( x ) = x3 − 6 x 2 + 9 x
The first order derivative of the function is f ( x ) = 3x 2 − 12 x + 9
The condition to find the stationary points is f ( x ) = 0 3x 2 − 12 x + 9 = 0
x = 1, 3 are the stationary points.
The second order derivative is f ( x ) = 6 x − 12
Implies f (1) = 6 (1) − 12 = −6 0
So, the stationary point x = 1 is the point of maximum, that maximize the function f ( x ) and
f ( 3) = 6 ( 3) − 12 = 6 0
The stationary point x = 3 is the point of minimum, that minimize the function f ( x ) .
DSSM Compiled by: Magisetty Harikrushna 3
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Construct the Hessian matrix for the function and check its definiteness
f ( X ) = 3e2 x1 +1 + 2e x2 +5
4 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Find the extreme value of f ( x, y ) = x3 + y 3 − 6 xy whether they are maximum or
minimum.
Solution: Given function is f ( x, y ) = x3 + y 3 − 6 xy
f f
Now = 3 x 2 − 6 y and = 3y2 − 6x
x y
To get stationary point we set these partial derivatives to zero
f
i.e., = 0 3x 2 − 6 y = 0
x
6 y = 3x 2
x2
y= → (1) and
2
f
= 0 3y2 − 6x = 0
y
3y2 = 6x
y 2 = 2 x → (2)
From (1) and (2)
2
x2
= 2x
2
x4
= 2x
4
x4 − 8x = 0
x ( x3 − 8 ) = 0
x = 0; x = 2
Implies y = 0; y = 2
Therefore, the stationary points are ( 0,0 ) ; ( 2, 2 )
2 f 2 f 2 f f
= 6 x; = 6 y ; − 6 and = −6
x 2
y 2
xy yx
Hessian matrix H ( X ) = 2 f ( X ) at point ( 0, 0 ) .
DSSM Compiled by: Magisetty Harikrushna 5
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
2 f 2 f
2
x xy 6 x −6 0 −6
H (X ) = 2 = =
f 2 f −6 6 y −6 0
yx y 2
Here H1 = 0 and H 2 = 36 , so the point is neither maxima nor minima.
Hessian matrix H ( X ) = 2 f ( X ) at point ( 2, 2 ) .
2 f 2 f
2
x xy 6 x −6 12 −6
H (X ) = 2 = =
f 2 f −6 6 y −6 12
yx y 2
Here H1 = 12 0 and H 2 = 108 0 ,
The matrix H is positive definite as all its leading principal minors are positive, so
So, the point ( 2, 2 ) is a point of minimum.
The minimum value is f ( 2 ) = ( 2 ) + ( 2 ) − 6 ( 2 )( 2 ) = −8
3 3
6 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Examine the extreme value of f ( X ) = x12 + x22 + x32 − 4 x1 − 8x2 − 12 x3 + 56 whether
they are maximum or minimum.
Problem: Construct the Hessian matrix, check its definiteness, and extremum values for the
function f ( X ) = 2 x12 + x22 + 3x32 + 10 x1 + 8 x2 + 6 x3 − 100
DSSM Compiled by: Magisetty Harikrushna 7
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Gradient descent method / Steepest Descent Method
Consider the objective function f ( X )
Minimize f ( X ) Set iteration number i = 1 start with X 1
Step 01. Find the search direction Si = −f ( X i ) is the gradient negative.
Step 02. Determine the optimal step length i in the direction of S i and set X i +1 = X i + i Si
SiT Si
Where i =
SiT HSi
Step 03. Test the optimality for the new point X i +1 by f ( X i +1 ) 0 .
If we reach the optimality stop, otherwise implement step 01 again for this new point X i +1.
Problem: Minimize by Gradient descent method f ( x1 , x2 ) = x1 − x2 + 2 x12 + 2 x1 x2 + x22 starting
0
from the point X 1 = .
0
Solution: The given objective function is f ( x1 , x2 ) = x1 − x2 + 2 x12 + 2 x1 x2 + x22
0
And starting from the point is X 1 = .
0
f f
Now = 1 + 4 x1 + 2 x2 and = −1 + 2 x1 + 2 x2
x1 x2
f
x 1 + 4 x + 2 x
The gradient of f is f = =
1 1 2
f −1 + 2 x1 + 2 x2
x
2
1 + 4 x1 + 2 x2
Implies −f = −
−1 + 2 x1 + 2 x2
2 f 2 f
2
x xy 4 2
Now the Hessian matrix is H ( X ) = 2 =
f 2 f 2 2
yx y 2
8 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
0
Iteration 01. At X 1 =
0
1
S1 = −f ( X1 ) = and
−1
SiT Si
We have i = T
Si HSi
S1T S1
1 = T
S1 HS1
−1
−1 1
1
1 =
4 2 −1
−1 1
2 2 1
2
1 = =1
2
Now X i +1 = X i + i Si
i.e., X 2 = X 1 + 1S1
0 −1 −1
X 2 = + (1) =
0 1 1
−1 0
We check the optimum f ( X 2 ) =
−1 0
So X 2 is not optimum, so move to next iteration.
−1
Iteration 02. At X 2 =
1
1
S2 = −f ( X 2 ) = and
1
SiT Si
We have i = T
Si H i S i
S2T S2
2 =
S2T HS2
DSSM Compiled by: Magisetty Harikrushna 9
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
1
1 1
2 = 1
4 2 1
1 1
2 2 1
2 1
2 = =
10 5
Now X i +1 = X i + i Si
i.e., X 3 = X 2 + 2 S2
−1 1 1 −0.8
X3 = + =
1 5 1 1.2
0.2 0
We check the optimum f ( X 3 ) =
−0.2 0
So X 3 is not optimum, so move to next iteration.
−0.8
Iteration 03. At X 3 =
1.2
−0.2
S3 = −f ( X 3 ) = and
0.2
SiT Si
We have i =
SiT H i Si
S3T S3
3 = T
S3 HS3
−0.2
−0.20.2
3 = 0.2
4 2 −0.2
−0.2 0.2
2 2 0.2
0.8
3 = =1
0.8
Now X i +1 = X i + i Si
i.e., X 4 = X 3 + 3 S3
−0.8 −0.2 −1.0
X4 = + (1) 0.2 = 1.4
1.2
10 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
−0.2 0
We check the optimum f ( X 4 ) =
−0.2 0
So X 4 is not optimum, so move to next iteration.
−1.0
Iteration 04. At X 4 =
1.4
0.2
S4 = −f ( X 4 ) = and
0.2
SiT Si
We have i =
SiT H i Si
S4T S4
4 =
S4T HS4
0.2
0.2
0.2
4 = 0.2
4 2 0.2
0.2 0.2
2 2 0.2
1
4 =
5
Now X i +1 = X i + i Si
i.e., X 5 = X 4 + 4 S4
−1.0 1 0.2 −0.96
X5 = + =
1.4 5 0.2 1.44
0.04 0
We check the optimum f ( X 5 ) =
−0.04 0
So X 5 is not optimum, so move to next iteration.
−0.96
Iteration 05. At X 5 =
1.44
−0.04
S5 = −f ( X 5 ) = and
0.04
SiT Si
We have i = T
Si H i S i
S5T S5
5 = T
S5 HS5
DSSM Compiled by: Magisetty Harikrushna 11
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
−0.04
−0.04
0.04
5 = 0.04
4 2 −0.04
−0.04 0.04
2 2 0.04
0.0032
5 = =1
0.0032
Now X i +1 = X i + i Si
i.e., X 6 = X 5 + 5 S5
−0.96 −0.04 −1
X6 = + (1) =
1.44 0.04 1.48
0.04 0
We check the optimum f ( X 5 ) =
0.04 0
−1.0
So X 6 = is the optimum,
1.48
Problem: Minimize by Steepest descent method f ( x1 , x2 ) = 4 x12 + 6 x22 − 8x1 x2 starting from the
1
point X 1 = .
1
Problem: Minimize by Gradient descent method f ( x1 , x2 ) = x12 + 2 x1 x2 + 2 x22 + x1 starting from
0.5
the point X 0 = .
0.5
12 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Gradient descent method to solve the following problem approximate sin ( x ) with a
degree 5 polynomial within the range −3 x 3 .
Solution: Let us consider the polynomial of degree 5 within the range −3 x 3 as
a0 + a1 x + a2 x2 + a3 x3 + a4 x4 + a5 x5
Let the loss function measures the error between the polynomial P ( x ) and sin ( x ) within the
range −3 x 3 .
sin ( x ) − P ( x )
2
The Mean Square Error (MSE) of the loss function is given by L = ,
N
where N is the set of data points in the given range.
Update the coefficients using grade descent method
If you calculate the value of each derivative, you will obtain the gradient for each
coefficient.
These are the values that give you the slope of the loss function with regards to each
specific coefficient.
They indicate whether you should increase or decrease it to reduce the loss, and also by
how much it should be safe to do so.
Given coefficients a0 , a1 , a2 , a3 , a4 , and a5 calculated gradients ga0 , ga1 , ga2 , ga3 , ga4 , ga5
and learning rate , typically one would update the coefficients so that their new, updated
values are defined as below:
a0 = a0 − ga0 ; a1 = a1 − ga1 ; a2 = a2 − ga2 ; a3 = a3 − ga3 ; a4 = a4 − ga4 ;
a5 = a5 − ga5
Repeat the above process and update the coefficients until, once you have applied that new
model to the data, your loss should have decreased.
Use these coefficients to get the polynomial of degree 5 within the given range.
DSSM Compiled by: Magisetty Harikrushna 13
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Constrained optimization
Karush-Kuhn-Tucker Conditions
The optimality conditions for a constrained local optimum are called the Karush Kuhn
Tucker (KKT) conditions and they play an important role in constrained optimization theory
and algorithm development. The KKT conditions for optimality are a set of necessary and
sufficient conditions for a solution to be optimal in a mathematical optimization problem.
The necessary and sufficient conditions
Consider the non-linear programming problem with inequality constraints
Maximize/Minimize f ( x )
Subject to gi ( x ) bi
The necessary and sufficient conditions for The necessary and sufficient conditions for
an absolute maximum of f ( x ) at x* are an absolute minimum of f ( x ) at x* are
f gi f gi
− i =0 − i =0
x i x x i x
i ( gi ( x) − bi ) = 0 i ( gi ( x) − bi ) = 0
g ( xi ) bi g ( xi ) bi
i 0 i 0
Minimize f ( x ) Minimize f ( x )
Subject to gi ( x ) bi Subject to gi ( x ) bi
f ( x ) is convex and g is convex f ( x ) is convex and g is concave
Maximize f ( x ) Maximize f ( x )
Subject to gi ( x ) bi Subject to gi ( x ) bi
f ( x ) is concave and g is convex f ( x ) is concave and g is concave
14 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Note:
Convex function always minimizes the problem. Concave function always maximizes
the problem.
A linear function is always convex function.
DSSM Compiled by: Magisetty Harikrushna 15
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Solve the following non-linear programming problem
Maximize Z = 8x1 + 10x2 − x12 − x22
Subject to 3x1 + 2 x2 6
x1 , x2 0
Solution: The given function is Maximize Z = 8x1 + 10x2 − x12 − x22
Subject to 3x1 + 2 x2 6
x1 , x2 0
Initially, we check the sufficient condition of Karush-Kuhn-Tucker Conditions
For the KKT conditions to be sufficient for Z to be maximum,
f ( x ) should be Concave and g ( x ) is convex.
Here f ( X ) = 8 x1 + 10 x2 − x12 − x22
f f
= 8 − 2 x1; = 10 − 2 x2
x1 x2
2 f 2 f
2
x xy −2 0
The Hessian matrix H ( X ) = 2 =
f 2 f 0 −2
yx y 2
Here H1 = −2 0 and H 2 = 4 0 ,
The matrix H is negative definite as all its leading principal minors are not zero and
alternate in sign, with first negative leading principal minor. So, H is negative definite and the
function f ( X ) is concave.
Clearly g ( x ) is convex as it is linear function.
Thus, the KKT conditions will be sufficient for maximum.
The Lagrangean function is L = f ( x ) − ( g ( x ) − b )
( )
i.e., L = 8x1 + 10 x2 − x12 − x22 − ( 3x1 + 2 x2 − 6 )
The necessary conditions are
L
= 0 ; g = 0 ; g ( x) b ; 0
X
16 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
L
= 8 − 2 x1 − 3 = 0 → (1)
x1
L
= 10 − 2 x2 − 2 = 0 → (2)
x2
( 3x1 + 2 x2 − 6 ) = 0 → (3)
0 → (4)
3x1 + 2 x2 6 → (5)
x1 , x2 0 → (6)
Case 01. When = 0
From (1) and (2)
x1 = 4; x2 = 5
But for these values the inequality (5) is not true.
Case 02. When 0
From (3) ( 3x1 + 2 x2 − 6 ) = 0
3x1 + 2 x2 − 6 = 0 → (7)
8 − 3
And from equation (1) 2 x1 = 8 − 3 x1 =
2
From equation (2) 2 x2 = 10 − 2 x2 = 5 −
Using x1 , x2 in equation (7)
8 − 3
3 + 2 (5 − ) − 6 = 0
2
32
= 0
13
8 − 3 4 33
Implies x1 = = and x2 = 5 − =
2 13 13
4 33 32
Hence the stationary point ( x1 , x2 ; ) = , , is the optimal Point.
13 13 13
2 2
4 33 4 33 277
And optimal value is Z = 8 + 10 − − = .
13 13 13 13 13
DSSM Compiled by: Magisetty Harikrushna 17
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Solve the following non-linear programming problem
Maximize Z = 3.6x1 − 0.4x12 + 1.6 x2 − 0.2 x22
Subject to 2 x1 + x2 10
x1 , x2 0
Solution: For 0
3.6 − 2 1.6 −
x1 = ; x2 = and
0.8 0.4
= 0.4
Implies x1 = 3.5; x2 = 3
Optimal value is Z = 10.7
Problem: Solve the following non-linear programming problem
Minimize Z = 10x1 + 4x2 − 2x12 − x22
Subject to 2 x1 + x2 5
x1 , x2 0
Solution: For 0
11 4
x1 = ; x2 = and
6 3
4
=
3
91
Optimal value is Z =
6
18 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Solve the following non-linear programming problem
Minimize Z = − log x1 − log x2
Subject to x1 + x2 2
x1 , x2 0
Solution: For 0
1 1
x1 = − ; x2 = − and
= −1
Implies x1 = 1; x2 = 1
Optimal value is Z = 0
DSSM Compiled by: Magisetty Harikrushna 19
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Maximize Z = 12x1 + 21x2 + 2x1 x2 − 2 x12 − 2 x22
Subject to x1 + x2 10; x2 8
x1 , x2 0
Solution: Given Maximize Z = 12x1 + 21x2 + 2x1 x2 − 2 x12 − 2 x22
Subject to x1 + x2 10; x2 8
x1 , x2 0
Initially, we check the sufficient condition of Karush-Kuhn-Tucker Conditions
For the KKT conditions to be sufficient for Z to be maximum,
f ( x ) should be Concave and g ( x ) is convex.
Here f ( X ) = 12 x1 + 21x2 + 2 x1 x2 − 2 x12 − 2 x22
f f
= 12 + 2 x2 − 4 x1; = 21 + 2 x1 − 4 x2
x1 x2
2 f 2 f
2
x xy −4 2
The Hessian matrix H ( X ) = 2 =
f 2 f 2 −4
yx y 2
Here H1 = −4 0 and H 2 = 12 0 ,
The matrix H is negative definite as all its leading principal minors are not zero and
alternate in sign, with first negative leading principal minor. So, H is negative definite and the
function f ( X ) is concave.
Clearly g1 ( x ) and g 2 ( x ) are convex functions as linear functions.
Thus, the KKT conditions will be sufficient for maximum.
The Lagrangean function is L = f ( x ) − 1 ( g1 ( x ) − b1 ) − 2 ( g 2 ( x ) − b2 )
( )
i.e., L = 12 x1 + 21x2 + 2 x1 x2 − 2 x12 − 2 x22 − 1 ( x1 + x2 − 10 ) − 2 ( x2 − 8)
The necessary conditions are
L
= 0 ; i ( gi − bi ) = 0 ; gi ( x ) bi ; i 0
X
L
= 12 + 2 x2 − 4 x1 − 1 = 0 → (1)
x1
20 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
L
= 21 + 2 x1 − 4 x2 − 1 − 2 = 0 → (2)
x2
1 ( x1 + x2 − 10 ) = 0 → (3)
2 ( x2 − 8) = 0 → (4)
1 , 2 0 → (5)
x1 + x2 10 → (6)
x2 8 → (7)
x1 , x2 0 → (8)
Case 01. 1 = 0; 2 = 0
From equations (1) and (2)
12 + 2 x2 − 4 x1 = 0
21 + 2 x1 − 4 x2 = 0
By solving these two equations we get
15
x1 = ; x2 = 9
2
These values do not satisfy the equations (6) and (7)
So, it is discarded
Case 02. 1 = 0; 2 0
From equation (4)
x2 − 8 = 0 x2 = 8
From equation (1)
12 + 2 ( 8) − 4 x1 − 0 = 0
x1 = 7
But the equation (6) is violated with these values
Thus, this case is also discarded.
Case 03. 1 0; 2 0
x1 + x2 − 10 = 0
DSSM Compiled by: Magisetty Harikrushna 21
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
x2 − 8 = 0 x2 = 8
And x1 = 2
Now from equation (1) and (2)
1 = 20; 2 = −27
Equation (5) is violated with these vales.
This case is also discarded.
Case 04. 1 0; 2 = 0
1 0 and form equation (3) we get x1 + x2 − 10 = 0 → ( A)
2 = 0 and from equations (1) and (2) we get
12 + 2 x2 − 4 x1 = 1
21 + 2 x1 − 4 x2 = 1
From above equations
12 + 2 x2 − 4 x1 = 21 + 2 x1 − 4 x2
Implies 6 x1 − 6 x2 + 9 = 0 → ( B)
From (A) and (B) x1 = 4.25; x2 = 5.75
In this case here 2 = 0 and
Form the equation 12 + 2 x2 − 4 x1 = 1
13
1 =
2
13
The optimal solution is ( x1 , x2 ; 1 , 2 ) = 4.25, 5.75; , 0
2
The optimal value is Z = 12 ( 4.25) + 21( 5.75) + 2 ( 4.25)( 5.75) − 2 ( 4.25) − 2 (5.75)
2 2
947
Z=
8
22 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Maximize Z = 4 x1 + 6 x2 − x12 − x22 − x32
Subject to x1 + x2 2; 2 x1 + 3x2 12
x1 , x2 0 ; x3 is unrestricted.
Solution:
H is negative definite and the function f ( X ) is concave.
Clearly g1 ( x ) and g 2 ( x ) are convex functions as linear functions.
Thus, the KKT conditions will be sufficient for maximum.
1 0; 2 = 0
1 3
x1 = ; x2 = ; x3 = 0; 1 = 3; 2 = 0
2 2
17
The optimal value is Z = .
2
DSSM Compiled by: Magisetty Harikrushna 23
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Minimize Z = ( x1 − 2 ) + ( x2 − 1)
2 2
Subject to x12 − x2 0; x1 + x2 2
x1 , x2 0
Solution: Given Minimize Z = ( x1 − 2 ) + ( x2 − 1)
2 2
Subject to x12 − x2 0; x1 + x2 2
x1 , x2 0
Initially, we check the sufficient condition of Karush-Kuhn-Tucker Conditions
For the KKT conditions to be sufficient for Z to be minimum,
f ( x ) should be Convex, g1 ( x ) and g 2 ( x ) should be Convex.
Here f ( X ) = ( x1 − 2 ) + ( x2 − 1)
2 2
f f
= 2 ( x1 − 2 ) ; = 2 ( x2 − 1)
x1 x2
2 f 2 f
2
x xy 2 0
The Hessian matrix H ( X ) = 2 =
f 2 f 0 2
yx y 2
Here H1 = 2 0 and H 2 = 4 0 ,
The matrix H is positive definite as all its leading principal minors are positive then function
f ( X ) is convex.
Here g1 ( x ) = x22 + ( − x1 ) is convex as sum of convex functions is convex.
Thus, the KKT conditions will be sufficient for minimum.
The Lagrangean function is L = f ( x ) − 1 ( g1 ( x ) − b1 ) − 2 ( g 2 ( x ) − b2 )
i.e., L = ( x1 − 2 ) + ( x2 − 1) − 1 ( x12 − x2 ) − 2 ( x1 + x2 − 2 )
2 2
The necessary conditions are
L
= 0 ; i ( gi − bi ) = 0 ; gi ( x ) bi ; i 0
X
L
= 2 ( x1 − 2 ) − 21 x1 − 2 = 0 → (1)
x1
24 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
L
= 2 ( x2 − 1) + 1 − 2 = 0 → (2)
x2
1 ( x12 − x2 ) = 0 → (3)
2 ( x1 + x2 − 2 ) = 0 → (4)
1 , 2 0 → (5)
x12 − x2 0 → (6)
x1 + x2 2 → (7)
x1 , x2 0 → (8)
Case 01. 1 = 0; 2 = 0
From equations (1) and (2)
2 ( x1 − 2 ) = 0 x1 = 2
2 ( x2 − 1) = 0 x2 = 1
These values do not satisfy the equations (6) and (7)
So, this case is discarded
Case 02. 1 = 0; 2 0
From equation (1), (2) and 1 = 0
2 ( x1 − 2 ) = 2
2 ( x2 − 1) = 2
Implies 2 ( x1 − 2 ) = 2 ( x2 − 1)
x1 − x2 = 1 → ( A)
From equation (4) and 2 0
x1 + x2 = 2 → ( B)
From (A) and (B)
1 1
x1 = and x2 = −
2 2
But the equation (8) is violated with these values.
Thus, this case is also discarded.
DSSM Compiled by: Magisetty Harikrushna 25
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Case 03. 1 0; 2 0
From equation (3)
x12 − x2 = 0 x12 = x2
From equation (4)
x1 + x2 − 2 = 0
x1 + x12 − 2 = 0
x1 = 1, − 2
As x1 0 x1 = 1 and
x2 = 1
Now from equation (1) and (2)
2 (1 − 2 ) − 21 − 2 = 0 21 + 2 + 2 = 0
2 (1 − 1) + 1 − 2 = 0 1 = 2
2
1 = 2 = − 0
3
2 2
The stationary point is ( x1 , x2 ; 1 , 2 ) = 1, 1; − , −
3 3
The optimal solution is Z = (1 − 2 ) + (1 − 1) = 1
2 2
26 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Solve a nonlinear programming problem by Karush-Kuhn-Tucker conditions
maximize f ( x, y ) = xy subject to x 2 + y 2 2; x, y 0 .
Solution: The given objective function is f ( x, y ) = xy
subject to x 2 + y 2 2; x, y 0
L = ( xy ) − ( x 2 + y 2 − 2)
The necessary conditions are
L
= 0 ; g = 0 ; g ( x) b ; 0
X
L
= y − 2 x = 0 → (1)
x
L
= x − 2 y = 0 → (2)
y
( x2 + y 2 − 2 ) = 0 → (3)
0 → (4)
x 2 + y 2 2 → (5)
x, y 0 → (6)
Case 01. When 0
From (3) x 2 + y 2 = 2
From (1) y = 2 x
From (2) x = 2 y
y = 2 ( 2 y )
y = 4 2 y
y − 4 2 y = 0
y (1 − 4 2 ) = 0
1 1
4 2 = 1 2 = =
4 2
1
Now =
2
DSSM Compiled by: Magisetty Harikrushna 27
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Implies x = y
From x 2 + y 2 = 2
x2 + x2 = 2
x2 = 1 x = 1
y =1
1
The stationary point ( x, y; ) = 1,1; is the optimal point
2
The optimal value is f ( x, y ) = (1)(1) = 1
28 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Solution of LPP by Graphical Method
Afterwards formulating the linear programming problem, our aim is to determine the
values of decision variables to find the optimum (maximum or minimum) value of the objective
function. Linear programming problems which involve only two variables can be solved by
graphical method. If the problem has three or more variables, the graphical method is
impractical.
The major steps involved in this method are as follows.
1. State the problem mathematically.
2. Write all the constraints in the form of equations and draw the graph.
3. Find the feasible region.
4. Find the coordinates of each vertex (corner points) of the feasible region. The coordinates
of the vertex can be obtained either by inspection or by solving the two equations of the
lines intersecting at the point.
5. By substituting these corner points in the objective function, we can get the values of the
objective function.
6. If the problem is maximization, then the maximum of the above values is the optimum
value. If the problem is minimization, then the minimum of the above values is the
optimum value.
DSSM Compiled by: Magisetty Harikrushna 29
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: Solve the following LPP
Maximize Z = 2 x1 + 5 x2
Subject to x1 + 4 x2 24
3x1 + x2 21
x1 + x2 9 and
x1 , x2 0
Solution: The given LPP is
Maximize Z = 2 x1 + 5 x2
Subject to x1 + 4 x2 24
3x1 + x2 21
x1 + x2 9 and
x1 , x2 0
First, we have to find the feasible region using the given conditions.
Since both the decision variables x1 and x2 are non-negative, the solution lies in the first
quadrant.
Write all the inequalities of the constraints in the form of equations.
x1 + 4 x2 = 24 ; 3x1 + x2 = 21 ; x1 + x2 = 9
Clearly
The line x1 + 4 x2 = 24 passing through ( 24, 0 ) ; ( 0, 6 )
x1 x2
+ =1
24 6
The line 3x1 + x2 = 21 passing through ( 7, 0 ) ; ( 0, 21)
x1 x2
+ =1
7 21
The line x1 + x2 = 9 passing through ( 9, 0 ) ; ( 0, 9 )
30 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
x1 x2
+ =1
9 9
DSSM Compiled by: Magisetty Harikrushna 31
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Now we draw the graph:
26
24
22
20
18
16
14
0 2 4 6 8 10 12 14 16
From the graph the shaded region is the feasible region, and the corner points of the feasible
region are O ( 0, 0 ) ; A ( 7, 0 ) ; B ( 6,3) ; C ( 4,5 ) ; D ( 0, 6 ) .
Objective function Corner Pont Function Value
O ( 0, 0 ) 0
A ( 7, 0 ) 14
Z = 2 x1 + 5 x2 B ( 6,3) 27
C ( 4,5 ) 33
D ( 0, 6 ) 30
Here the optimum point is C ( 4,5 ) and the optimum value is 33.
32 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Solve the following LPP
Maximize Z = 2 x1 + 3x2
Subject to x1 + x2 30
x2 12
x1 20 and
x1 , x2 0
Problem: Solve the following LPP
Minimize Z = 5 x1 + 4 x2
Subject to 4 x1 + x2 40
2 x1 + 3 x2 90 and
x1 , x2 0
DSSM Compiled by: Magisetty Harikrushna 33
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A company makes two products (X and Y) using two machines (A and B). Each unit
of X that is produced requires 50 minutes processing time on machine A and 30 minutes
processing time on machine B. Each unit of Y that is produced requires 24 minutes processing
time on machine A and 33 minutes processing time on machine B. At the start of the current
week there are 30 units of X and 90 units of Y in stock. Available processing time on machine
A is forecast to be 40 hours and on machine B is forecast to be 35 hours. The demand for X in
the current week is forecast to be 75 units and for Y is forecast to be 95 units. Company policy
is to maximize the combined sum of the units of X and the units of Y in stock at the end of the
week. (i) Formulate the problem of deciding how much of each product to make in the current
week as a linear program. (ii) Solve this linear program graphically.
Solution:
Let x be the no. of product X and y be the no. of product Y.
The no. of product X and the no. of product Y are our decision variables.
The constraints are the available processing time on the machines. The available processing
time on machine A is forecast to be 40 hours. Since it is in hours whereas it is given in minutes
for processing X and Y, we need to convert hours into minutes.
Therefore, 50 x + 24 y 2400
Similarly, the available processing time on machine B is 30 x + 33 y 2100
The total demand for products X and Y in the current week is forecasted as 75 units and 95
units respectively. To maximise the no. of units of X and no. of units of Y left in stock at the
end of the week,
Z = ( x + 30 − 75) + ( y + 90 − 95)
i.e., Z = x + y − 50 is the objective function.
The demand for X and Y are 75 units and 95 units respectively whereas initially, they are in
stock of 30 units of X and 90 units of Y.
The non-negativity constraints,
x 75 − 30 x 45 and y 95 − 90 y 5
Hence, on formulating the linear program,
34 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Maximize, Z = x + y − 50
Subject to, 50 x + 24 y 2400
30 x + 33 y 2100
x 45
y5
We write these inequalities in to equalities & plot these equalities on a graph as shown below.
Here the line 50 x + 24 y 2400 pass through the points ( 48, 0 ) and ( 0, 100 )
line 30 x + 33 y 2100 pass through the points ( 70, 0 ) and ( 0, 63.63) ,
The lines y = 5 and x = 45 are parallel to X-axis and Y-axis respectively.
DSSM Compiled by: Magisetty Harikrushna 35
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Graph:
100
95
90
85
80
75
70
65
60 v
55
50
45
40 v v
35
30
25
20
15
10
05
0 05 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Here, we obtain the corner points are A ( 45, 6.25) , B ( 45,5 ) and C ( 45.6,5 ) .
Objective function Corner Pont Function Value
A ( 45, 6.25 ) 1.25
Z = x + y − 50 B ( 45,5 ) 0
C ( 45.6,5 ) 0.6
36 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Here, the maximum sum obtained is for point A ( 45, 6.25) . Hence, no. of units of X should
be 45 and no. of units of Y should be 6.25 to obtain the maximum sum of products in the
current week.
Number of units of X should be 45 and number of units of Y should be 6.25 to obtain the
maximum sum of products in the current week.
DSSM Compiled by: Magisetty Harikrushna 37
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Problem: A company produces two types of bulbs A and B. bulbs A is of superior quality and
bulbs B is of lower quality. Profits on bulbs A and B are Rs 15 and Rs 10 per bulbs respectively.
Raw materials required for each bulb A is twice as that of bulb B. The supply of raw material
is sufficient only for 1000 bulbs per day. bulb A requires a special clip and only 400 such clips
are available per day. For bulbs B, only 700 clips are available per day. Formulate this problem
as a linear programming problem, and solve it graphically.
Solution: Let x and y denotes the number of bulbs in type A and type B respectively
Objective function:
Profit on x bulbs in type A is 15x
Profit on y bulbs in type B is 10 y
Total profit is 15 x + 10 y
Profit is to be maximized, therefore, the objective function is
Maximize Z = 15 x + 10 y
Constraints:
Raw materials required for each bulb A is twice as that of bulb B.
i.e., for bulb A raw material required is 2x and for B is y .
Raw material is sufficient only for 1000 bulb per day
i.e., 2 x + y 1000
Bulb A requires 400 clips per day
i.e., x 400
Bulb B requires 700 clips per day
i.e., y 700
Non-negative restrictions:
Since the number of bulbs is non-negative,
we have x 0; y 0
Thus, the mathematical formulation of the LPP is
Maximize Z = 15 x + 10 y
Subject to 2 x + y 1000
x 400
y 700 and
x 0; y 0
38 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: A company manufactures two models of electric Kettles viz., ordinary and auto-cut.
All components of the Kettles are purchased from outside sources, assembly and testing is
carried out at company’s own works. The assembly and testing time required for the two
models are 0.8 hour each for ordinary and 1.20 hours each for auto-cut. Manufacturing capacity
720 hours at present is available per week. The market for the two models has been surveyed
which suggests maximum weekly sale of 600 units of ordinary and 400 units of auto-cut. Profit
per unit for ordinary and auto-cut models has been estimated at Rs 100 and Rs 150 respectively.
Formulate the linear programming problem, and solve it graphically.
Solution: Let x1 and x2 denotes the number of electric Kettles of two types ordinary and auto-
cut, respectively.
Objective function:
Profit on x1 ordinary Kettles is 100x1
Profit on x2 auto-cut Kettles is 150x2
Total profit is 100 x1 + 150 x2
Profit is to be maximized, therefore, the objective function is
Maximize Z = 100 x1 + 150 x2
Constraints:
The assembling and testing time required for x1 units of ordinary Kettles is 0.8x1 and for x2
units of auto-cut Kettles is 1.2x2
Since the manufacturing capacity is 720 hours per week.
We get 0.8 x1 + 1.2 x2 720
maximum weekly sale of ordinary Kettle is 600
i.e., x1 600
maximum weekly sale of auto-cut Kettle is 400
i.e., x2 400
Non-negative restrictions:
Since the number of both types of Kettle s is non-negative,
we have x1 0; x2 0
Thus, the mathematical formulation of the LPP is
DSSM Compiled by: Magisetty Harikrushna 39
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Maximize Z = 100 x1 + 150 x2
Subject to 0.8 x1 + 1.2 x2 720
x1 600
x2 400 and
x1 0; x2 0
Writing the inequalities into equalities we get,
0.8 x1 + 1.2 x2 = 720 this line is passing through ( 900, 0 ) and ( 0, 600 )
x1 = 600 and x2 = 400
Graphical solution:
1000 950
950
900
850
800
750
700
650
600
550
500
450
400
350
300
250
200
150
100
50
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000
950
40 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Problem: Determine the slope and intercept and graph the linear inequalities:
i) 5 x + 2 y −6
ii) 8 x − 5 y 5
Solution: The given linear inequalities 5 x + 2 y −6; 8 x − 5 y 5
Writing in to equalities 5 x + 2 y = −6 and 8 x − 5 y = 5
5 x + 2 y = −6 2 y = −5 x − 6
5
y = − x − 3 is in the form y = mx + c
2
−5
Here m = is the slope, Y- intercept c = −3 .
2
8x − 5 y = 5 5 y = 8x − 5
8
y= x − 1 is in the form y = mx + c
5
8
Here m = is the slope, Y- intercept c = −1 .
5
−6
The line 5 x + 2 y = −6 passes through the points , 0 and ( 0, −3) .
5
5
The line 8 x − 5 y = 5 passes through the points , 0 and ( 0, −1) .
8
Graph
2
-3 -2 -1 0
1
DSSM Compiled by: Magisetty Harikrushna 41
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Introduction to least squares optimization
Least squares optimization and constrained optimization techniques, which are
extensively used to analyse and visualize data. Least squares (LS) optimization problems are
those in which the objective (error) function is a quadratic function of the parameter(s) being
optimized. The solutions to such problems may be computed analytically using the tools of
linear algebra, have a natural and intuitive interpretation in terms of distances in Euclidean
geometry, and also have a statistical interpretation.
Optimization in ML
In fact, behind every Machine Learning (and Deep Learning) algorithm, some optimization is
involved. let us take the simplest possible example. Consider the following 3 data points
X Y
1 4.6
3 12.8
6 15.5
Everybody conversant with machine learning will immediately recognize that we are referring
to X as the independent variable (also called “Features” or “Attributes”), and the Y is the
dependent variable (also referred to as the “Target” or “Outcome”). Hence, the overall task of
any machine is to find the relationship between X & Y. This relationship is
actually “Learned” by the machine from the DATA, and hence we call the term Machine
Learning. We, humans, learn from our experiences, similarly, the same experience is fed into
the machine in the form of data.
Now, let us say that we want to find the best fit line through the above 3 data points. The
following plot shows these 3 data points in blue circles. Also shown is the green line, which
we are claiming as the “Best-Fit Line” through these 3 data points. Also, have shown a “Poor-
Fitting” line (the red line) for comparison.
42 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
Y
Original Data
16 Poor Fit Line
Best Fit Line
14
12
10
08
06
04
02
X
0.0 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
The main objective is to find the Equation of the Best-Fitting Straight Line through these three
data points.
Y = a0 + a1 X → (1) is the equation of the best-fit line green-line in the above plot, where a1
slope of the line, a0 intercept of the line. In machine learning, this best fit is also called
the Linear Regression (LR) model, and w0 and w1 are also called model weights or model
coefficients.
The predicted values from the Linear Regression model (Y) are represented by orange-squares
in the above plot. Obviously, the predicted values are NOT exactly the same as the actual values
of Y (blue circles), the vertical difference represents the error in the prediction given by
Error = Yi − Y i → (2) for any ith data point. Now we claim that this best fit-line will have the
minimum error for prediction, among all possible infinite random “poor-fit” lines. This total
Error across all the data points is expressed as the Mean Squared Error (MSE) Function, which
will be minimum for the best-fit line.
1 n
(
Yi − Y i )
2
MSE = → (3) , where ‘n’ is the total number of data points in the dataset.
n i =1
DSSM Compiled by: Magisetty Harikrushna 43
Reference Material MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.)
Minimizing or maximizing any quantity is mathematically referred to as an Optimization
Problem, and hence the solution (the point where the minimum/maximum exists) is referred
the optimal values of the variables.
There are two types of Supervised learning tasks, Regression & Classification.
Regression
In Regression problems, the objective is to find the “best-fit” line passing through the
majority of the data points and will have the minimum value of an error function, also called
as the Cost function or Loss function, commonly which is the Mean Square Error (MSE). Why
only MSE, why not a simple sum of errors terms or mean the absolute error will be discussed
in another article. Hence, as discussed in the above section, the main objective is to find the
line having the minimum value of this error function.
The following table lists the Error functions being minimized in the most commonly used
regressor algorithms.
Name of the Regression Algorithm Loss (Cost) Function minimized
Linear Regression Mean Square Error (MSE)
Polynomial Regression Mean Square Error (MSE)
Regularized Regression (Ridge/LASSO) SSE + Penalty Term
Decision Tree Regressor (DTR) Sum of Squares of Errors (SSE)
L1 − Loss (Same as sum of absolute values
Support Vector Regressor (SVR)
of errors) Or L2 − Loss (Similar to SSE)
44 Compiled by: Magisetty Harikrushna DSSM
MAPTSMS; MIAENG; MAMS; MIAEPT M.Sc., B.Ed., M.Phil., (Ph.D.) Reference Material
You might also like
- Dimensionality Reduction Using PCA (Principal Component Analysis)No ratings yetDimensionality Reduction Using PCA (Principal Component Analysis)13 pages
- National Institute of Technology, Tiruchirappalli MBA Trimester Examination, Basic Data Analytic Marathon ExamNo ratings yetNational Institute of Technology, Tiruchirappalli MBA Trimester Examination, Basic Data Analytic Marathon Exam22 pages
- FALLSEM2023-24 - ITE2011 - ETH - VL2023240102356 - 2023-09-01 - Reference-Material-I (3 Files Merged)No ratings yetFALLSEM2023-24 - ITE2011 - ETH - VL2023240102356 - 2023-09-01 - Reference-Material-I (3 Files Merged)191 pages
- 11668a5f867641748200d0bfd6a889a3_hst951_7No ratings yet11668a5f867641748200d0bfd6a889a3_hst951_732 pages
- Projecting Data To A Lower Dimension With PCANo ratings yetProjecting Data To A Lower Dimension With PCA6 pages
- Principal Component Analysis: Courtesy:University of Louisville, CVIP LabNo ratings yetPrincipal Component Analysis: Courtesy:University of Louisville, CVIP Lab48 pages
- A NGUNYI NOTES SMA 2332 Probability and Statistics IVNo ratings yetA NGUNYI NOTES SMA 2332 Probability and Statistics IV64 pages
- Important Matrices For Multivariate AnalysisNo ratings yetImportant Matrices For Multivariate Analysis7 pages
- 1.12.2024-BSC-301-CSBS-class note_2024-25No ratings yet1.12.2024-BSC-301-CSBS-class note_2024-2558 pages
- Multiple Regression Model - Matrix FormNo ratings yetMultiple Regression Model - Matrix Form22 pages
- How Do You Do A Principal Component Analysis?No ratings yetHow Do You Do A Principal Component Analysis?13 pages
- Methods of Multivariate Statistics: Study Guide ForNo ratings yetMethods of Multivariate Statistics: Study Guide For219 pages
- Probability Distributions and Curve FittingNo ratings yetProbability Distributions and Curve Fitting53 pages
- Linear Regression Analysis: Smoking and Lung CapacityNo ratings yetLinear Regression Analysis: Smoking and Lung Capacity16 pages
- 3 - Noman Naseer - Dimentionality - Reduction and Principle Component AnalysisNo ratings yet3 - Noman Naseer - Dimentionality - Reduction and Principle Component Analysis43 pages
- Module 3-Correlation and Regression: X y and X yNo ratings yetModule 3-Correlation and Regression: X y and X y74 pages
- Mathematics III (Prob&Stats) - Unit 4 5No ratings yetMathematics III (Prob&Stats) - Unit 4 5122 pages
- Factoring and Algebra - A Selection of Classic Mathematical Articles Containing Examples and Exercises on the Subject of Algebra (Mathematics Series)From EverandFactoring and Algebra - A Selection of Classic Mathematical Articles Containing Examples and Exercises on the Subject of Algebra (Mathematics Series)No ratings yet
- Learn Statistics Fast: A Simplified Detailed Version for StudentsFrom EverandLearn Statistics Fast: A Simplified Detailed Version for StudentsNo ratings yet
- Linear Systems and Control - Week 10: Observer Based State Feedback Controller DesignNo ratings yetLinear Systems and Control - Week 10: Observer Based State Feedback Controller Design37 pages
- Introduction to Environmental Data Science Hsieh instant downloadNo ratings yetIntroduction to Environmental Data Science Hsieh instant download23 pages
- Design Formulas For Tuned Mass Dampers Based O N A Perturbation TechniqueNo ratings yetDesign Formulas For Tuned Mass Dampers Based O N A Perturbation Technique22 pages
- Instant download Mathematica by Example 5th Edition Martha L. Abell & James P. Braselton pdf all chapter100% (4)Instant download Mathematica by Example 5th Edition Martha L. Abell & James P. Braselton pdf all chapter41 pages
- Dunkerley'S Method of Approximation To Findout The Natural Frequency of Multi Degree Freedom SystemNo ratings yetDunkerley'S Method of Approximation To Findout The Natural Frequency of Multi Degree Freedom System9 pages
- Identifikasi Huruf Kapital Tulisan Tangan Menggunakan: Linear Discriminant Analysis Dan Euclidean DistanceNo ratings yetIdentifikasi Huruf Kapital Tulisan Tangan Menggunakan: Linear Discriminant Analysis Dan Euclidean Distance11 pages
- Stata Graph Library For Network AnalysisNo ratings yetStata Graph Library For Network Analysis41 pages
- M2A1: Double Taylor Series & The Jacobian Matrix 1 Double Taylor SeriesNo ratings yetM2A1: Double Taylor Series & The Jacobian Matrix 1 Double Taylor Series2 pages
- Get (Ebook) Foundations of Factor Analysis, Second Edition by Stanley A Mulaik ISBN 9781420099614, 1420099612 PDF ebook with Full Chapters Now100% (4)Get (Ebook) Foundations of Factor Analysis, Second Edition by Stanley A Mulaik ISBN 9781420099614, 1420099612 PDF ebook with Full Chapters Now86 pages
- Immediate download Matrix Analysis and Applications 1st Edition Xian-Da Zhang ebooks 2024100% (3)Immediate download Matrix Analysis and Applications 1st Edition Xian-Da Zhang ebooks 202435 pages
