확률 및 통계(Probability & Engineering Statistics)

Multiple Random Variables

5.1 Introduction

 5.1 Introduction


◈ Chap1~4 : single random variable defined on a given sample space

◈ Chap 5 : multivariate systems (2 variable  bivariate random variables)
◈ Single random variable은 random한 문제를 다루는 powerful tool이지만, 보통 공학적인 문제는 2차원 이상의
다차원 문제가 많으므로 single R.V.을 확장해야 함.
예) 궁수가 화살로 과녁 맞추기
- Fortunately, many situations of interest in engineering can be handled by the theory of two R.V.
- 시간영역으로 확장하면, autocorrelation, cross correlation, covariance 등과 밀접하게 연관됨.
◈ Random Variable  Multiple Random Variable (random Vector)  Random Process

 5.2 Joint CDFs of Bivariate Random Variables (or Vector Random Variables)


◈ Consider two random variables X and Y defined on the same sample space
- 예) X : 학생들의 성적, Y : 같은 학생들의 키

◈ Joint CDF(결합확률분포함수) of bivariate random variables

FX ,Y ( x, y ) = P[ X ≤ x, Y ≤ y ] FX ( x) = P[ X ≤ x]
Marginal CDF
FY ( x) = P[Y ≤ y ]
◈ X와 Y가 독립인 경우
FX ,Y ( x, y ) = FX ( x) FY ( y )
 Multiple Random Variables

5.2 Joint CDFs of Bivariate Random Variables

 Vector Random Variable

◈ Mapping from the sample space S to the joint sample space SJ

Joint sample space (결합표본공간)

~ range sample space (치역표본공간)
~ 2-dimensional product space (2차원곱공간)

 Mapping 함수가
2개임을 기억할 것!

◈ Event A와 Event B를 정의하자.

A = { X ≤ x}
B = {Y ≤ y}

Joint Event A ∩ B = { X ≤ x and Y ≤ y}

◈ N차원으로 확장 가능하다. (우선은 2차원에만 집중하자!)

 Multiple Random Variables

5.2 Joint CDFs of Bivariate Random Variables

 Joint Distribution Function (결합분포함수)

◈ Event A and B

A = { X ≤ x} FX ( x) = P{ X ≤ x}
B = {Y ≤ y} FY ( y ) = P{Y ≤ y}

◈ The probability of the joint event -> joint probability distribution function (결합확률분포함수)

Joint Event A ∩ B = { X ≤ x and Y ≤ y} FX ,Y ( x, y ) = P{ X ≤ x and Y ≤ y} = P( A ∩ B)

◈ Example)
- SJ (Joint Sample Space) : (1,1), (2,1), and (3,3)
- Probability : P(1,1)=0.2, P(2,1)=0.3, and P(3,3)=0.5
Find FX,Y(x,y)
 Multiple Random Variables

5.2 Joint CDFs of Bivariate Random Variables

 Joint Distribution
◈ Example) A fair coin is tossed twice.  S={(H,H), (H,T), (T,H), (T,T)}, 확률 { ¼, ¼, ¼, ¼ }
X = “number of heads on the first toss”, Y=“number of heads on the second toss”
- mapping : Heads를 1로, Tail을 0으로

- PDF

- CDF 계산

Independent
- marginal pdf
f X ( x) = 2 / 4δ ( x) + 2 / 4δ ( x − 1)
f Y ( y ) = 2 / 4δ ( y ) + 2 / 4δ ( y − 1)
 Multiple Random Variables

5.2 Joint CDFs of Bivariate Random Variables

 Properties of the Joint Distribution Function

◈ 일차인 경우와 비교하여 설명하면,
: 공집합과의 교집합은 zero이므로

(1) Fx ( −∞) = 0 (4,5) FX ,Y (−∞, ∞) = 0, FX ,Y (−∞, y ) = 0, FX ,Y ( x,−∞) = 0

(2) Fx (∞) = 1 (3) FX ,Y (∞, ∞) = 1
(3) 0 ≤ Fx ( x) ≤ 1 (1) 0 ≤ FX ,Y ( x, y ) ≤ 1
(4) Fx ( x1 ) ≤ Fx ( x2 ) if x1 ≤ x2
(2) FX ,Y ( x, y ) is a nondecreasing function of both x and y
(5) P{x1 ≤ X ≤ x2 } = Fx ( x2 ) − Fx ( x1 )
(8) P{x1 < X ≤ x2 , Y ≤ y} = FX ,Y ( x2 , y ) − FX ,Y ( x1 , y )
(6) Fx ( x + ) = Fx ( x)
(9) P{ X ≤ x, y1 < Y ≤ Y2 } = FX ,Y ( x, y 2 ) − FX ,Y ( x, y1 )

(10) P{x1 < X ≤ x2 , y1 < Y ≤ Y2 } = FX ,Y ( x2 , y 2 ) − FX ,Y ( x1 , y 2 ) − FX ,Y ( x2 , y1 ) + FX ,Y ( x1 , y1 )

(6) lim+ FX ,Y ( x, y ) = FX ,Y (a, y ) : 앞에 2개에서
x→a 중복되어 뺐으므로
(7) lim+ FX ,Y ( x, y ) = FX ,Y ( x, b) 더해줌(교집합 고려)
y →b FX ,Y ( x, ∞) = FX ( x), FX ,Y (∞, y ) = FY ( y ) marginal distribution functions 개념

Sample space S 전체이므로 S와 x의 교집합은 x 만 남음

◈ Example 5.1) 그림을 그려서 이해할 것! P{ X > a and Y > b} = P{ X > a ∩ Y > b}
= 1 − P{ X ≤ a ∪ Y ≤ b}
= 1 − FX ( x) − FY ( y ) + FX ,Y ( x, y )
 Multiple Random Variables

5.3 Discrete Random Variables

 Discrete Random Variables

◈ Joint PMF (or Joint PDF) : 결합 밀도함수 p X ,Y ( x, y ) = P[ X = x, Y = y ]

◈ 결합밀도함수의 특성
1) PMF는 negative가 될 수도 1보다 클 수도 없음 0 ≤ p X ,Y ( x, y ) ≤ 1
2) 결합밀도함수의 총합은 1

∑∑ p
x y
X ,Y ( x, y ) = 1

3) 결합밀도함수를 a와 b까지 더한 값은 CDF가 됨 ∑∑ p

x ≤ a y ≤b
X ,Y ( x, y ) = FXY (a, b)

◈ Example 5.2) 결합밀도함수가 다음과 같을 때 k값은?

p X ,Y ( x, y ) = k (2 x + y ) x = 1,2; y = 1,2 4k
6k
- 18k=1이어야 하므로 k=1/18
3k
- marginal PMF는?
pX(x) = {7/18 at x=1,11/18, at x=2} 5k
pY(y) = {8/18 at y=1,10/18, at y=2}
- 독립인가?
x=1, y=1일 때, pXY(x,y)=3/18, pX(x) = 7/18 , pY(y) = 8/18이므로 두 값이 곱과 다르므로 dependent

◈ Example 5.4) 그림을 그려서 이해할 것!

 Multiple Random Variables

5.3 Discrete Random Variables

 Discrete Random Variables

◈ Example) Discrete Random Variable N M
- SJ (Joint Sample Space) : (1,1), (2,1), and (3,3) FX ,Y ( x, y ) = ∑∑ P( x , y
n =1 m =1
n m )u ( x − x n )u ( y − ym )
- Probability : P(1,1)=0.2, P(2,1)=0.3, and P(3,3)=0.5
Find FX,Y(x,y)

N M
FX ,Y ( x, y ) = ∑∑ P( x , y
n =1 m =1
n m )u ( x − x n )u ( y − y m ) = P (1,1)u ( x − 1)u ( y − 1) + P(2,1)u ( x − 2)u ( y − 1) + P(3,3)u ( x − 3)u ( y − 3)

N M
f X ,Y ( x, y ) = ∑∑ P( x , y
n =1 m =1
n m )δ ( x − x n )δ ( y − y m ) = P(1,1)δ ( x − 1)δ ( y − 1) + P(2,1)δ ( x − 2)δ ( y − 1) + P(3,3)δ ( x − 3)δ ( y − 3)
 Multiple Random Variables

5.3 Discrete Random Variables

 Marginal Distribution Functions (부분분포함수)

◈ The distribution function of one R.V. can be obtained by setting the value of the other variable to infinity.

◈ Example) 앞의 예제 에서
N M
FX ,Y ( x, y ) = ∑∑ P( x , y
n =1 m =1
n m )u ( x − x n )u ( y − ym )

= P (1,1)u ( x − 1)u ( y − 1) + P (2,1)u ( x − 2)u ( y − 1) + P(3,3)u ( x − 3)u ( y − 3)

FX ( x) = FX ,Y ( x, ∞) = P(1,1)u ( x − 1) + P(2,1)u ( x − 2) + P (3,3)u ( x − 3)

FY ( y ) = FX ,Y (∞, y ) = P(1,1)u ( y − 1) + P(2,1)u ( y − 1) + P(3,3)u ( y − 3)

= 0.2u ( y − 1) + 0.3u ( y − 1) + 0.5u ( y − 3) = 0.5u ( y − 1) + 0.5u ( y − 3)
 Multiple Random Variables

5.4 Continuous Random Variables

 Joint Density Function (결합밀도함수)

◈ Joint density is defined by the 2nd derivative of the joint distribution function

∂ 2 FX ,Y ( x, y )
f X ,Y ( x, y ) = Joint Density Function N차 미분까지 개념 확장 가능
∂x∂y

N M
f X ,Y ( x, y ) = ∑∑ P( x , y
n =1 m =1
n m )δ ( x − x n )δ ( y − ym ) : 불연속인 경우 delta함수를 이용하여 정의

◈ Joint Density Function의 특성

x ∞

(1) f X ,Y ( x, y ) ≥ 0 ∫ ∫
( 4) F X ( x ) =
−∞ −∞
f X ,Y ( x, y )dxdy
y ∞
∞ ∞ F ( y) = ∫ ∫ f X ,Y ( x, y )dxdy
∫ ∫ f X ,Y ( x, y )dxdy = 1 Y
(2) −∞ −∞
−∞ −∞
x y
(3) FX ,Y ( x, y ) = ∫ ∫
−∞ −∞
f X ,Y ( x, y )dxdy
x2 y2
∫ ∫
∞
(5) P{x1 < x ≤ x2 , y1 < y ≤ y 2 } =
x1 y1
f X ,Y ( x, y )dxdy (6) f X ( x) = ∫ −∞
f X ,Y ( x, y )dy
∞
fY ( y) = ∫ f X ,Y ( x, y )dx
−∞
 Multiple Random Variables

5.4 Continuous Random Variables

 Joint Density Function

◈ Example) b가 양수일 때 다음 함수가 pdf가 되려면?

be − x cos( y ), 0 ≤ x ≤ 2 and 0 ≤ y ≤ π / 2

g ( x, y ) = 
 0, all other x and y

(1) f X ,Y ( x, y ) ≥ 0
∞ ∞
(2) ∫ ∫
−∞ −∞
f X ,Y ( x, y )dxdy = 1

π /2 2 2 π /2
∫ ∫
0 0
be − x cos( y )dxdy = b ∫
0 ∫
e − x dx cos( y )dy
0 1
−2
= b(1 − e ) = 1

b = 1 /(1 − e −2 )
 Multiple Random Variables

5.4 Continuous Random Variables

 Marginal Probability Density Function (부분확률밀도함수)

◈ Marginal probability density function

∞
( 6) f X ( x ) = ∫ −∞
f X ,Y ( x, y )dy
∞
fY ( y) = ∫ f X ,Y ( x, y )dx
−∞

Marginal pdf

◈ Example 5.5) 다음 함수로부터 marginal PDF를 구하고, 이를 통해 independent인지 보이시오.

f X ,Y ( x, y ) = u ( x)u ( y )e − ( x + y )

∞ ∞
f X ( x) = ∫ −∞
f X ,Y ( x, y )dy = u ( x)e − x ∫ 0
e − y dy = u ( x)e − x

∞ ∞
f Y ( x) = ∫ −∞
f X ,Y ( x, y )dx = u ( y )e − y ∫ 0
e − x dx = u ( y )e − y
 Multiple Random Variables

5.4 Continuous Random Variables

 Statistical Independence
◈ 확률에서 통계적 독립이란?

P( A | B) = P( A), P( B | A) = P( B) -> When events are independent, it will often be found

P( A ∩ B) = P( A) P( B) that probability problems are greatly simplified.

◈ Random Variable에서는

P{ X ≤ x, Y ≤ y} = P{ X ≤ x}P{Y ≤ y}
pdf와 cdf에 대해서 정리하면 FX ,Y ( x, y ) = FX ( x) FY ( y ), f X ,Y ( x, y ) = f X ( x) f Y ( y )

◈ Example) 다음 함수의 경우는 “not independent”

∞ ∞
f X ( x) = ∫ −∞
f X ,Y ( x, y )dy = u ( x) xe − x ∫0
e − xy dy = u ( x)e − x
− x ( y +1)
f X ,Y ( x, y ) = u ( x)u ( y ) xe ≠ ∞ ∞
∫ ∫
u( y)
f Y ( x) = f X ,Y ( x, y )dx = u ( y ) xe − x e − xy dx =
−∞ 0 ( y + 1) 2
◈ Example) 다음 함수의 경우는 “independent”
∞ ∞

1
f X ( x) = ∫ −∞
f X ,Y ( x, y )dy = u ( x)e −( x / 4) ∫0
e −( y / 3) dy = u ( x)e −( x / 4)
f X ,Y ( x, y ) = u ( x)u ( y )e −( x / 4) −( y / 3) = ∞ ∞
12 fY ( y) = ∫ f X ,Y ( x, y )dyx = u ( x)e −( y / 3) ∫ e −( x / 4) dyx = u ( x)e −( y / 3)
−∞ 0
 Multiple Random Variables

5.5 Determining Probabilities from a Joint CDF

 5.5 Determining Probabilities from Joint CDF

◈ Example 5.7) 그림을 그려서 이해할 것!

 Multiple Random Variables

5.6 Conditional Distributions

 5.6.1 Conditional PMF for Discrete Random Variables

◈ 조건 확률

P( A ∩ B) The probability of an event A may depend on a second event B

P( A | B) =
P( B) If A and B are mutually exclusive, P(A|B)=0

◈ Conditional PMF

P[ X = x, Y = y ] p XY ( x, y ) P[ X = x, Y = y ] p XY ( x, y )
pY | X ( y | x) = = p X |Y ( x | y ) = =
P[ X = x] p X ( x) P[Y = y ] pY ( y )

1
◈ Example 5.8) p X ,Y ( x, y ) = (2 x + y ) x = 1,2; y = 1,2
18 4/18
6/18
2

∑ 18 (2 x + y) = 18 ∑ (2 x + y) = 18 (4 x + 3)
1 1 1
p X ( x) = x = 1,2 3/18
y y =1
5/18
2

∑ ∑
1 1 1
pY ( x) = (2 x + y ) = (2 x + y ) = (2 y + 6) y = 1,2
x
18 18 x =1
18

P[ X = x, Y = y ] p XY ( x, y ) 2 x + y
pY | X ( y | x) = = =
P[ X = x] p X ( x) 4x + 3
 Multiple Random Variables

5.6 Conditional Distributions

 5.6.2 Conditional PDF for Continuous Random Variables

◈ Conditional PDF

f XY ( x, y ) f XY ( x, y )
f Y | X ( y | x) = f X |Y ( x | y ) =
f X ( x) fY ( y)
◈ Example 5.9)
 xe − x ( y +1) , 0 ≤ x ≤ ∞; 0 ≤ y ≤ ∞
f XY ( x, y ) = 
 0, otherwise
∞
∞ ∞  e − xy 
∫ ∫
−x − xy −x −x
f X ( x) = f X ,Y ( x, y )dy = xe e dy = xe −  =e
−∞ 0
 x  0
∞ ∞ e − x ( y +1)
f Y (yx) = ∫−∞
f X ,Y ( x, y )dx = ∫0
xe − x ( y +1) dx dv = e − x ( y +1) dx, v = −
y +1
부분적분 필요 u=x

∞ ∞
xe − x ( y +1) ∞ e − x ( y +1) e − x ( y +1) ∞
∫ ∫
1
E[ X ] = [uv]∞0 − v du = [− ]0 + dx = 0 + [− ] =
2 0
y +1 y +1 ( y + 1) ( y + 1) 2
0 0

f XY ( x, y ) xe − x ( y +1) f XY ( x, y )
f Y | X ( y | x) = = −x
= e − xy f X |Y ( x | y ) = = x( y + 1) 2 e − x ( y +1)
f X ( x) e fY ( y)
 Multiple Random Variables

5.6 Conditional Distributions

 5.6.3 Conditional Means and Variances

◈ Conditional mean : µY | X = E[Y | X = x]

◈ Conditional variance : σ Y2| X = E[(Y − µY | X ) 2 | X = x] = E[Y 2 | X = x] − (E[Y | X = x])2

◈ Example 5.10) : conditional mean E[X:Y=y]를 구하시오.

 e −( x / y ) e − y
 , 0 ≤ x ≤ ∞; 0 ≤ y ≤ ∞
f XY ( x, y ) =  y
 0,
 otherwise

∞ ∞ e−x / ye− y e− y ∞
f Y ( x) = ∫
−∞
f X ,Y ( x, y )dx = ∫0 y
dx =
y ∫ 0
e − x / y dx =e − y

f XY ( x, y ) e − ( x / y ) e − y e − ( x / y )
f X |Y ( x | y ) = = =
fY ( y) ye − y y

∞ ∞
∫ ∫
1
E[ X | Y = y ] = xf X |Y ( x | y )dx = xe − x / y dx
0 y 0

부분적분 필요 u = x, dv = e − x / y dx → du = dx, v = − ye − x / y

1
[
 − xye
y
−x / y
]∞
0 +y ∫ 0
∞ 
e − x / y dx  = y
→ y
 Multiple Random Variables

5.6 Conditional Distributions

 5.6.4 Simple Rule for Independence

◈ X와 Y가 독립인 경우
 X와 Y의 Joint PDF가 직각좌표계에서 다음과 같은 형태를 가질 경우 독립이다.
f XY ( x, y ) = constant × x − factor × y − factor , a ≤ x ≤ b, c ≤ y ≤ d

◈ Example 5.12) 다음 Joint PDF는 독립인가?

1 3
f XY ( x, y ) = x y, 0 ≤ x ≤ 2, 0 ≤ y ≤ 1
2
1
f XY ( x, y ) = f X ( x) × f Y ( y ) = x 3 y, 0 ≤ x ≤ 2, 0 ≤ y ≤ 1
2
 Multiple Random Variables

5.7 Covariance and Correlation Coefficient

 5.7 Covariance and Correlation Coefficient

◈ 두 개의 R.V. X와 Y의 covariance의 정의
Cor ( X , Y ) = σ XY = E[( X − µ X )(Y − µY )]
= E[ XY − µY X − µ X Y + µ X µY ]
= E[ XY ] − µ X µY X와 Y가 독립이면, Covariance는 zero가 됨

◈ Correlation coefficient
 a measure of how good a prediction of the value of one of the two random variables cab be formed
based on an observed value of the other.
Cor ( X , Y ) σ XY
ρ XY = =
Var ( X )Var (Y ) σ XσY -1에서 1까지의 값을 가짐

 correlation이 +와 –의 의미는?

◈ Example 5.13) 독립인지 확인 필요

25e −5 y , 0 ≤ x ≤ 0.2; 0 ≤ y ≤ ∞
f XY ( x, y ) = 
 0, otherwise
 Multiple Random Variables

5.7 Covariance and Correlation Coefficient

 5.7 Covariance and Correlation Coefficient

◈ Example 5.14) 5분 기다리다 감

- Hans’s arrival time vs. Ann’s arrival time
f X ( x) = 1 / 60, 0 ≤ x ≤ 60 f Y ( y ) = 1 / 30, 15 ≤ x ≤ 45

- Probability that they meet - Probability that Ann arrives before Hans
 P[ X − Y ≤ 5]  P[Y ≤ X ]

Ann
Y
X
Y=
45
A: Hans가 10분 이전에 도착하는 경우
5

D: Hans가 50분 이후에 도착하는 경우

=
X
-

B : Hans가 10~40분 사이에 도착하고

Y

30 A B E C D
Ann이 5분 후에 도착하는 경우
5
=
Y

C : Hans가 20~50분 사이에 도착하고

-
X

15 Ann이 5분 전에 도착하는 경우

30
5
0 X Hans
-5 5 10 20 50 60