¢ University under wtida'3 of UGC Ack 1986)

eee OF ADVANCED SCIENCES

DEPARTMENT OF MATHEMATICS
FALL SEMESTER - 2020~2021

MAT2001 - Statistics for Engineers

(Embedded Theory Component)

COURSE MATERIAL

Module 4
Probability Distributions

Syllabus:
Binomial Distribution - Poisson Distribution - Normal Distribution -
Gamma Distribution - Exponential Distribution - Weibull Distribution.

Prepared By: Prof. K. Kavitha (In-charge)

Prof. M. Gowsalya
Prof. Rajesh Moharana

The course in-charges thankfully acknowledge the course materials preparation

committee in-charge and members for their significant contribution in bringing
out of this course material.

eK RK

Dr. D. Easwaramoorthy

Dr. A. Manimaran

Course In-charges - MAT2001-SE,

Fall Semester 2020~2021,
Department of Mathematics,

SAS, VIT, Vellore.

Module 4. Special distribution

Discrete distributions

Continuous distributions

1. Binomial distribution

1.Uniform distribution (or)

Rectangular distribution

2. Poisson distribution

2.Exponential distribution

3. Geometric distribution

3.Gamma distribution

4.Negative binomial 4.Weibull distribution

distribution

5.Normal distribution
Properties:

1. There must be a fixed number of trials.

2. All trials must have identical probabilities of success

(P).

3. The trials must be independent of each other.

Notations:

n= number of trials

p = probability of success

q = probability of failure

X= A random variable which represents the number

of successes.
Applications:

Sampling process (or) Quality control measures in

industries to classify the items as defective or non-
defective.

Binomial distribution:

Probability mass function of Binomial distribution:

A discrete random variable X is said to follow binomial

distribution if its probability mass function is

P(x) =N. Pa”, =O yack

Moment generating function:

The moment generating function of a binomial variate is

M,(0) = Efe” |

=(Pe' +q)”
Mean and Variance of Binomial distribution:
l d {
fy = GT {M .(t)} <0

d ‘
= [{(pe' +q) }Jr<0

n-l

=[n(pe’ +4)"~ pe" ],-0

=n(p+q)"'p [ez p+q=l]
=np

Mean of the Binomial distribution is = np

] 2
uy =| 73 {M.(0)}],<0

at

d?

=F {(pe' +q)"} ] 0

n—-l

/
=[— tn(pe' +q)"" pe! }],00

-l ot
n-l 4

= np[e'(n—1)(pe' +q)""? pe’ +(pe’ +q) Imo

=npl(n-l)(p+q)"* p+(p+q)""|] [ev ptq=l

=np|(n—-l)p+l|
by =n(n— l)p? +np
Variance of Binomial distribution:

My = My — (ay)
=n(n—1)p?> +np —n’ p°
=n’ p> —np*+np —n’p°’
= np(1— p)

npg

Variance of Binomial distribution is npq

Problems:

1. The mean of a binomial distribution is 5 and standard

deviation is 2. Determine the distribution.

Given mean= np = 5
S.D =V"Pqd =
npq= 4
5q =4
q= 4/5
p = 1-q=1/5
We have np=5
n(1/5) =5
n=25

Hence the Binomial distribution is

P(X =x)=N¢ p'q”*=25¢(D*)*.

2. The mean and variance of a binomial distribution are
4 and 4/3. Find P(X 21)

Given mean np=4

Variance npq = 4/3

npq i oT

np 4 3
qg=1/3 ; p=2/3

np =4
n(2/3) =4
n =6

p(X >1)=1-P(X <1)

=]— P(X =0)

=1-6¢ pq”
= 1-6, (2/3)"(1/3)°°

=1-—(1/3)° =0.998
3. An irregular 6 faced die is such that the probability
that it gives 3 even numbers in 5 throws is twice the
probability that it gives 2 even numbers in 5 throws.
How many sets of exactly 5 trials can be expected to
give no even number out of 2500 sets?

Let X denote the number of even numbers obtained

in 5 trias.

Given P(x=3)=2 *P(x=2)

Scie * Zs
Sc, P79 =2"*5c¢, Pq

p=2q=2(1- p)=2-2p
tp = 2

ie., Pp = 2 /3
g=I1/'S
Now, P(getting no even number)=P(x=0)
|
5c pg =(1/3 =—
cP =(/3) 343

Number of sets having no success (no even number) out

of N sets=N *P(x=0)

Required number of sets=2500* 1/243=10, nearly.

Poisson distribution

A discrete random variable X is said to follow Poisson

distribution with parameter 4 if its probability mass

—A 4x
P(x) == < ~ x=012...000 on

function is
Poisson distribution as limiting form of Binomial
distribution under the following conditions

i) n, the number of trials is indefinitely large,

len,

ii) p, the constant probability of success in each trial

is very small. ie., p> 0.

lii) np (=2) is finite (or) p=4/n and q=1-4/n,

where 2 is a positive real number.

The following are some of the examples, which may be

analysed using Poisson distribution.
1.The number of alpha particles emitted by a
radioactive source in a given time interval.
2.The number of telephone calls received at a

telephone exchange in a given time interval.

3. The number of defective articles in a packet of 100.
4.The number of printing errors at each page of a
book.
5. The number of road accidents reported in a city
per day.
Moment generating function:

The MGF of the Poisson variate X with parameter 7 is

given as

M, (t) =e") = Ye™ p(x)

x=0

oO e” y

= e*

x!
( Ae ie

x!

. 0
—@* Sy
x=0

oO
4) Aut a
=e age ° —e"

— Ae —l)
Mean and variance of Poisson distribution

1_|d
fy = —M y(t)
dt ‘=

Mean of Poisson distribution = A

My ~ : . uw. |
dt ~
~ t=0
= ee fete)
dt dt 1=0
= d (e*e** 20") |
| dt an
= Ae *(e** 2! 4 efe** Ze] ,9
= Ae*[e* 4 e@7A]
=e *e* «Ae 472"
fa =A* +A
Variance of Poisson distribution

H2 = Ly ~ (a)?

=V+A-V =A
In Poisson distribution Mean = Variance =2
Problems:

1. The number of monthly breakdowns of a computer

is a random variable having a Poisson distribution
with mean equal to 1.8. Find the probability that
this computer will function for a month.

a) Without a breakdown
b) With only one breakdown and
c) With atleast one breakdown

Let X denotes the number of breakdowns of the

computer in a month.

X follows a Poisson distribution with mean 2 =1.8

—-A 4x

P(X =x)= etn

x!

e t8 (1.8)

x!
a) p(x =0)= e '* = 0.1653
b) p(x =1) =e | 3(1.8) = 0.2975
c) p(x 21) =1- p(x = 0) = 0.8347
2.1f the probability of a defective fuse from a
manufacturing unit is 2%, in a box of 200 fuses, find the

probability that a) exactly 4 fuses are defective b) more

than 3 fuses are defective.

P=0.02, n=200
Mean =4 =np=200*0.02=4

—j, 5x
a) P(X =4)=——
x!
hay
- £4)" _ 9.1952

b) P(x >3)=1-—P(x $3)

=1—[p(x=3) + p(x =2)+ p(x =1)+ p(x =9)]
a 2 1 0
=) eg 5 595600
ee
Fitting of Binomial Distribution

Fitting a binomial distribution means assuming that the given

distribution is approximately binomial and hence finding the
probability mass function and then finding the theoretical

frequencies.

Problem:

Fit a binomial distribution for the following data:

x 0 1 2 3 4 5 6 Total
f 5 18 28 12 7 6 + 80
Xx f fx Theoretical
frequency
0 5 0 4
1 18 18 15
2 28 56 25
3 12 36 22
4 7 28 11
5 6 30 3
6 4 24 0
>f=N=80 f= 192
z= fr 192 _
— Lf 80

np =2.4(or)6p=2.4

p=0.4,

q = 0.6
Theoretical distribution is given by

=Nx P(X = x)
6
=80 x » 6. pq" *
x=0

= 80x q° +6, q°'p +6. gr se Pisses

3

=80x[(0.6)° +6, (0.6)°"'(0.4) + 6c, (0.6)°*(0.4)” + 6,, (0.6)°*(0.4)

+ 6,,(0.6)°-4(0.4)" +6, (0.6)°-*(0.4)* +6, (0.4)°]

x: D 1 2 3 4 5 6
fe 3:73 14.93 24.88 22.12 11.06 2.95 0.33

Fitting of Poisson Distribution

1. Fit a Poisson distribution to the following data and

calculate the theoretical frequencies.

Deaths 0} 1) 2 |3/4
Frequency | 122 | 60 | 15 | 2 {1
x f fx Theoretical
frequency
0 122 0 121
1 60 60 61
2 15 30 15
3 2 6 3
4 1 4 0
N=200 | } fr=100
Exponential Distribution:
Definition: A continuous random variable X is said to follow an
exponential distribution with parameter 2 >0, if its probability density

function is given by

f@la)=2e7*%, x>0,4 >0.

It is also known as negative exponential distribution.

e Mean of the exponential distribution is 1 / a

e Variance of the exponential distribution is 1 / 42°

e The moment generating function of the exponential distribution

is*/, _4 ,a>t.

e Ther” moment about the origin is "he ,r =1,2,3,

Examples related to exponential distribution:

Example:
The mileage which car owners get with a certain kind of radial tire is a RV

having an exponential distribution with mean 40,000 km. Find the probabilities
that one of these tires will last (i) at least 20,000 km and (ii) at most 30,000
km.
Solution:

_ Let X denote the mileage obtained with the tire

_ 41 -x/40,000
fx) = 40.000 € x>0
* l
(i) P(X = 20,000) =f ——— ¢*40.000 ay
20,000 40,000
= —x/ 40,000 ]"°
= [-e lesen
=e = 0.6065
: 30,000 4 iim
ii) P(X < 30,000) = meen A
(ii) P¢ ) ) 0,000 e
* _ [- e7 */ 40,000 70,000

“= 1-¢°7=0,5270
Exercise:

The time (in hours) required to repair a machine is exponentially distributed with
parameter A = 1/2.
(a) What is the probability that the repair time exceeds 2 h?
(b) What is the conditional probability that a repair takes at least 10 h given
that its duration exceeds 2h?

FANT:

S) PCX> 9)

(by. P(x os 9 —PO>I0,x>4)

eee

} (x>4)

Normal Distribution:

Definition: A continuous random variable X is said to have a normal

distribution if its probability function is given by

ov2r

eee
F(x) = e 2c —oo <x < 00, —co<p<oo, a >0,
where y and o are the mean and standard deviation of the normal random

variable X, respectively.

If X follows normal distribution, then it is denoted by X~ N(u,c). It is also

called as the Gaussian distribution.

Standard Normal Distribution:

eye . X-p . .
Definition: The standard normal random variable Z = — is said to have a

standard normal distribution if its probability function is defined by

A standard normal variate is denoted by N(0,1). The standard normal

distribution is also known as a Z-distribution or a unit normal distribution.
Standardization of a normal distribution helps us to make use of the tables of the

area of the standard curve.

Distribution function of a standard normal variable:

The distribution function @(z) of a standard normal variate, Z is defined

by

o(z)=P(Z<z) = / f(t)dt

1 res 1,2
= ————s / e zt dt.
V20 J —ce

Properties of @(z):
@ o(-—z) =1- 9(z)
@ P(a<X <b) = (34) — 024)
@ P(iZ<a)=P(Z=>

—a)
Graph of normal distribution:

Inflection points
—S

p-30 po2exmle uw wtle n+20 at 3e

oe

96.46%

———_- 27

Area under normal curve:

The curve of a normal distribution is bell shaped, with the highest point
over the mean jz. It is symmetrical about a vertical line through jz. The
normal curve has the following features.

@ The curve is symmetrical about the coordinate at its mean which

locates the peak of the bell.

@ The values of mean, median and mode are equal.

@ The curve extends from —oo to oo.

@ The area covered between js — 0 and 44 +c is 0.6826 (i.e. 68.26 %

of area).

@ The area covered between js — 20 and js + 20 is 0.9544 (i.e. 95.44

% of area).

@ The area covered between js — 30 and ys + 30 is 0.9974 (i.e. 99.74

% of area).
Zaz

The following while gives the shaded area in the diagram viz. P(O<Z <2)

for different vabucs of z.

Mar

3
\} “we <xrce Area cneet Nora! Curve

TABLE OF AREAS

£-

~ N(O, 1)

AREAS UNDER NORMAL CURVE

Noemat probability curve is given by

on aoe {- 3

oe) Feo ( - bs), erse

and standard normal probatuity curve is given by

where Z

~| REESE SRES PASTS FUGEE THES TEE GE TERT E EES

~| SESSE Eaees RHEE SERGE SHETE GEE EE TERE ERE
> Gqage SLRS SERIE TRSGE LEGGE REESE SEES EEES
«| @E88E RARE SERSE TSHGE SITET THEE ST EEEE EEE
~| SEGRE ERAGR SEZTE FRESE ELGEE TSE GE G ETE LEER

| E8ESE BAERS EREEG RPEGTERGEE GREE EGECE CEES

o| BREESE REAR HERES STRTE TICES ESTEE EETEE EGER
«| QEG83 BATES TAREE RTGS CREST gERTE THERE EET
~| EEGRS ERTS NGG POETS CREES TREE EEGEE ETE
PEELS Rakag OSNGE GgRgECGERE ETETETETEE ETE

§lertey cores ssees Bete Lanz agers LaK43 A2a8

Area under standard normal curve:

The standard normal random variable Z and the normal random variable
X have identical curves.

Qo

oO

The standard normal curve covers 68.26 % area between z = —1

and z= 1. Therefore, P(—1 < Z < 1) = 0.6826.
The standard normal curve covers 95.44 % area between z = —2
and z = 2. Therefore, P(—2 < Z < 2) = 0.9544.
The standard normal curve covers 99.74 % area between z = —3

and z =3. Therefore, P(—3 < Z < 3) = 0.9974.

Properties of a normal curve:

In a normal distribution, mean deviation about mean is approximately

4. . ie
equal to 5 limes its standard deviation.

In a normal distribution, the quartiles Q, and Q3 are equidistant from the

median.

The normal curve is bell shaped and is symmetrical about X = wu.

The area bound by the normal curve and the x-axis equal to 1.

The tails of the curve of a normal distribution extend identically in both

sides of x = yu and never touch the x-axis.

The moment generating function of a normal distribution with respect to

1,2,2
woe t+=t?o
origi 18 ef! 2 .
Examples related to normal distribution:

Example:

The marks obtained by a number of students in a certain subject are approxi-

mately normally distributed with mean 65 and Standard deviation 5, If 3 students
are’selected at random from this group, what is the probability that at least | of
them would have scored above 757

Solution:

If X represents the marks obtained by the students, X follows the distribution

N65, 5).
P(a student scores above 75)

75-65 X-65 )
-

——— <

= PX > 75) = P( 5

= P(2 <Z <>), (where Z is the standard normal variate)

= 0.5 -— P(0<Z< 2)

= 0.5 — 0.4772, (from the table of areas)

= 0.0228
Let p = P(a student scores above 75) = 0.0228 then g = 0.9772 and n = 3.
Since p is.the same for all the students, the number Y, of (successes) students

scoring above 75, follows a binomial distribution.

P(at least 1 student s scores above 75)

= P(at least 1 success)

= P(Y2 l= 1— P(Y¥=0)

= 1—nC, x Pq"

= 1 ~3C, (0.9772)

= 1-—0.9333

= 0.0667

Example:

In an engineering examination, a student is considered to have failed, secured

second class, first class and distinction, according as he scores less than 45%,
between 45% and 60%, between 60% and 75% and above 75% respectively. In
a particular year 10% of the students failed in the examination and 5% of the
students got distinction. Find the percentages of students who have got first class
and second class. (Assume norma! distribution of marks).
Solution:

LetX represent the percentage of marks scored by the students in the examina- A
tion.

Let X follow the distribution N (44, 7). rT

. Given: P (X < 45) = 0.10 and P(X > 75) =0.05 Z

ie., P(-=< tet nad )} =0.10 ana

Co Co

r( zB <X=ten)=005 NG OQ Wx AY

oem
. o(n<e Droit yo EK Fy
P =e = 0.05 45<\J <t5

p(o<z<#=*3).= 0.40 and

, P(o< z< 3-H) = 0.45
Prom the inble of areas, we get
AS. 128 and BEE = 164
ie., u=-1.28 o= 45 (1)
and p+ 1.64 0=75 . (2)

Solving equations (1) and (2), we get

M= 58.15 and o= 10.28

Now P (a student gets first class) .

= P(60 < X < 75)
=P{ 60 — 58.15 eze 75 — 58.15
10.28 19.28

= P(0.18 <Z < 1.64}

=P{0<Z< 1%4}~ P{0<Z<0.18}

= 0.4495 — 0.0714 = 0.3781

.. Percentage of students getting first class = 38 (approximately)
Now percentage of students getting second class

= 100 — (sum of the percentages of students who have failed,

got first class and got distinction)

= 100 — (10 + 38 + 5), approximately.

= 47 (approximately)
WEIBULL DISTRIBUTION

Definition: A continuous RV X is said to follow a Weibull distribution with

parameters a, B> 0, ifthe RV Y=aX ® follows the exponential distribution with
density function f(y) = e”, y > 0.

Density Function of the Weibull Distribution

Since ¥ = a- XP, we have y =a: x?.
By the transformation rule, derived in chapter 3, we have f,(x) = fy( y) | a .
x
where f(x) and f(y) are the density functions of X and Y respectively.
fx) = €? aBxP!
=apx’-! es x>0 (~ y> 0]

Mean and Variance of Weibull Distribution

The raw moments 4,” about the origin of the Weibull distribution are given by
H,’ = E(X")

= ap { yi tB-1 owe oy
0

ee eich 2
af (ZF er (Z)F wy,
a a

on putting y= ax? orx= (2)?

- a" | y Per dy
0

=a" ig:
é +1)
aL
Mean = E(X) = 44’ = & (ee)

Var(X) = E(X*) — (£00)?

GHG]
In a certain city, the daily consumption of electric power in millions of kilowatt-
hours can be treated as a RV having an Erlang distribution with parameters

A= + andk=3. Ifthe power plant of this city has a daily capacity of 12 millions

kilowatt-hours, what is the probability that this power supply will be inadequate
on any given day.

Let X represent the daily consumption of electric power (in millions of kilo-
watt-hours). Then the density function of X is given as

L 3
pow) 2 oe x>0

P (the power supply is inadequate)

- t
= P(X>12)= f fix)dx [+ The daily capacity is only 12]
n

[Gy
1 2 ev eu ear
"mi boll ee a
2 4 8

= oe (288 + 96 + 16)
= 25 ¢* = 0.0625
If a company employes n sales persons, its gross sales in thousands of rupees
may be regarded as a RV having an Erlang distribution with A = ; and k =
80 Jn. If the sales cost is Rs. 8000 per salesperson, how many salespersons
should the company employ to maximise the expected profit?

Let X represent the gross sales (in Rupees) by n salespersons,

X follows the Erlang distribution with parameters A= + and k = 80,000 Jn.

BOO = ; = 1,60,000 Jn

If y denotes the total expected profit of the company, then

y = total expected sales —total sales cost

= 1,60,000 /n — 8000 »

dy . 80,000
dn n

= 0, when Jn = 10 or n= 100

2
<> = - S099 <0, when n = 100. ’

Therefore, y is maximum, when n= 100.

That is the company should employ 100 salespersons in order to maximise the
total expected profit.
Consumer demand for milk in a certain locality, per month, is known to be a
general Gamma (Erlang) RV. If the average demand is a litres and the. most
likely demand is b litres (b < a), what is the variance of the demand?

Let X represent the monthly consumer demand of milk.

Average demand is the valucof E(X).

Most likely demand is the value of the mode of X or the value of X for which
its density function is maximum,

If f (x) is the density function of X, then

k
fe) = ote 230

at —2 .-Axr 1 Ax
FG) = pee (-t ett Ata) er}
= ae tte 1) Ash
= 0, when x= 0, x= <—*
&
r@= paras (k= 1) = Ax) (tN)

k-1
<0, when x = ——
““h

Therefore f(x) is maximum, when x = + .

i.e., Most likely demand = “= =b (1)

and BU) = =a (2)
k_ kil

Vv =—.—

Now wit) ra
=a(a—b), {from (1) and (2)}
Each of the 6 tubes of a radio set has a life length (in years) which may be
considered as a RV that follows a Weibull distribution with parameters a = 25
and B= 2. If these tubes function independently of one another, what is the prob-
ability that no tube will have to be replaced during the first 2 months of service?

If X represents the life length of each tube, then its density function f(x) is
given by

f(x) = aBxP-' @ x>0

ie, f(x) = 50x e*™* x>0
Now P (a tube is not to be replaced during the first 2 months)

=P(x>+)

50x eo" dx

oF

= (-e7"), = 625

. P(all the 6 tubes are not to be replaced during the first 2 months)
= (e7)* (by independence)
=¢

= 0.0155
If the time T to failure of a component follows a Weibull distribution with para-
meters @ and , find the hazard rate or conditional failure rate at time ¢ of the
component.

Refer to Example (19) in Worked Example 2(A).

If (2) is the density function of T and A(1) is the hazard rate at time f, then

f()
1— F(t)

where F(z) is the distribution function of T.

Now. f= afxr’-'e@™ 1>0

FO) = PT Ss)

A(O=

at apr", a
ae

[7]

a)}-¢"

p-1 .-at?
n(n = SPE

= apr?-!
If the life X (in years) of a certain type of car has a Weibull distribution with
the

parameter B= 2, find the value of the parameter cr, given that probability that the
' life of the car exceeds 5 years ise~°**. For these values of a and f, find the
mean

and variance of X.
The density function of X is given by

f(x) =2axe,x>0 [+ B=2)

Now P(X>5) = j 2ax e * dx

5

atusnery
je],
=e
Given that lp ee
Pa ~ 0.25

6 eal
~ Ragtcomenn (to) *((3)
= tox +{(4)

veon- ch fF
(5) ‘[ro-{(@¥]
-['-(4)|

x
100 (1 - 2)
Gamma and Weibull Distribution
GAMMA DISTRIBUTION:

Definition: A continuous RV X is said to follow an Erlang distribution or

General Gamma distribution with parameters 4 > 9 and k > 0, eee

Samir eaerats giventy

k k=l .=Ax
f@)= Axe” for x>0

(k)
0, otherwise

T ae ee ses
Waianae | rater |X e* de

molt t*-! edt, [on putting Ax =]

0

=]
Hence f (x) is a legitimate density function.

Mean and Variance of Erlang Distribution

The raw meet H,’ about the origin of the Eiiang distribution are given by
"= E(X’)

~—™

= f a Pe oe
5 1h)
oe te ced T yweves e -
7 eed t dt, (on putting Ax =
_ 1 |k+r)
WV tk
. me (k +1) _*k
Mean = E(x) = 4 (k) A
Var(X) = E(X*) E00)
= 4, We _(4y
# [ke A

= oy (kk +1)- 2) =