1 of 45

EXPECTED VALUE OF A FUNCTION &

JOINT MGF

9:01

Prepared by : M. Dokhi, Ph.D.

Sessions target
2 of 45

Nilai harapan bersama (Joint expected value)

Kovarian (Covariance)
Korelasi (Correlation)
Nilai Harapan Bersyarat (Conditional Expected Value)
MGF Bersama (Joint MGF)

9:01

Prepared by : M. Dokhi, Ph.D.

Expected value (Review)

3 of 45
Expected value of discrete R.V.

f ( x) p x (1 p)1 x ; x 0,1
1

E ( X ) x p x (1 p )1 x 0 p 0 (1 p )1 1 p1 (1 p )0 p
x 0
1

E ( X 2 ) x 2 p x (1 p )1 x 02 p 0 (1 p)1 12 p1 (1 p) 0 p
x 0

V ( X ) E[ X ] E[ X ] p p 2 p(1 p)
2

9:01

Prepared by : M. Dokhi, Ph.D.

Expected value (Review)

4 of 45

( 1)

Expected value of continuous R.V.

f ( x)

1
e
2

1
x2
2

; x

x
k a

xe

x e x dx

dx k !a k 1

12

1 12 x2
1 12 x2
E( X ) x
e dx 2 x
e dx 0

0
2
2

E( X 2 ) x2

1
e
2

1
x2
2

dx 2 x 2
0

V ( X ) E[ X 2 ] E[ X ] 1 02 1
2

9:01

Prepared by : M. Dokhi, Ph.D.

1
e
2

1
x2
2

dx 1

Joint pdf and expected value

5 of 45

Let X and Y be random variables with joint probability f(x, y).

Their expected values (means) are written as

Discrete random variables:

X x f ( x, y )
x

Y y f ( x, y )

X x f ( x)

or

Y y f ( y )
y

Continuous random variables:

x f ( x, y)dydx

9:01

Prepared by : M. Dokhi, Ph.D.

x f ( x)dx

y f ( x, y)dxdy

or

y f ( y)dy

Contoh
6 of 45
A joint pdf of two random variables X, Y is given by
xy
96
f ( x, y )
0

for

E[ X ]

0 x 4, 1 y 5

xf ( x, y )dxdy

otherwise

E[Y ]

xy
x
dydx
96

x 0 y 1
4

x2 y
dydx
96

x2 y2

192
0
4

dx

1
5

x2
dx
8

x3 4 8

24 3
9:01
0
Prepared by : M. Dokhi, Ph.D.

yf ( x, y)dxdy

xy
y
x0 y1 96 dydx
4

xy 2
dydx

96

0 1
4
3 5
xy

dx

288
0
1
4

31x
dx
72
0

31x 2

144

31

9
0
4

Expected value of a function

7 of 45

R.V. X

f ( x)

Misalkan H u( X ) adalah sembarang fungsi dari X, maka

EH

u ( x). f ( x)

; X diskrit

u( x) f ( x)dx

; X kontinu

Jika u ( X ) X E[u ( X )] E[ X ] Rata-rata X X

Jika u( X ) ( X E[ X ])2 E u ( X ) E X E ( X )

E( X 2 ) E[ X ] V X X2
2

9:01

Prepared by : M. Dokhi, Ph.D.

Expected value of a function

8 of 45

If X and Y has a joint pdf f(x,y) and if H u ( X , Y ) is a

function of X and Y, then

Discrete random variables:

E( H ) u( x, y) f ( x, y)
x

Continuous random variables:

E ( H ) ... u( x, y) f ( x, y)dxdy

9:01

Prepared by : M. Dokhi, Ph.D.

Expected value of a function

9 of 45

If f ( X 1 , X 2 ,..., X k ) adalah joint fungsi dari X 1 , X 2 ,..., X k

maka untuk H u ( X 1 , X 2 ,..., X k )

E ( H ) ... u( x1 ,..., xk ) f ( x1 ,..., xk )

x1

xk

E ( H ) ... u( x1 ,..., xk ) f ( x1 ,..., xk )dx1...dxk

x1

9:01

xk

Prepared by : M. Dokhi, Ph.D.

Jika X 1 , X 2 ,..., X k diskrit

Jika X 1 , X 2 ,..., X k kontinu

Example
10 of 45

Joint pdf of two random variables X, Y is given by

xy
96
f ( x, y )
0

0 x 4, 1 y 5

for

otherwise

Let H u ( X , Y ) 2 X 3Y
The expected value of H is

E[2 X 3Y ]

(2 x 3 y) f ( x, y)dxdy

xy
(2 x 3 y) dxdy
96
x 0 y 1
4

x 2 y xy 2

dydx
48 32
0 1
47

3
4

9:01

Prepared by : M. Dokhi, Ph.D.

Covariance: Definition
11 of 45

Misalkan H u( X , Y ) ( X E[ X ])(Y E[Y ])

E H E X E[ X ]Y E[Y ]

E X X Y Y

X X Y Y f ( X , Y )

X,Y discrete

X X Y Y f ( X , Y )dxdy

X,Y continuous

y x

Covar X , Y
XY
9:01

Prepared by : M. Dokhi, Ph.D.

Ukuran keeratan
hubungan linear
antara 2 R.V.

Variance vs. Covariance

12 of 45
Case of 2 R.V.

Case of 1 R.V.

Var X X2

Covar X , Y XY
E X E[ X ]Y E[Y ]

E X E( X )

E ( X 2 ) E[ X ]

Rumus hitung
Variance

9:01
Prepared by : M. Dokhi, Ph.D.

E ( XY ) E[ X ]E[Y ]

Rumus hitung
Covariance

Var( X ) Covar( X , X )

Some properties of Covariance

13 of 45

If X and Y are random variables and a and

b are constant, then

Cov(aX , bY ) abCov( X , Y )
Cov(a X , b Y ) Cov( X , Y )
Cov( X , aX b) aCov( X , X ) aVar( X )

If X and Y are independent, then

Cov( X , Y ) E( XY ) E( X ).E(Y ) 0
9:01

Prepared by : M. Dokhi, Ph.D.

Example

Prepared by : M. Dokhi, Ph.D.

Example

Prepared by : M. Dokhi, Ph.D.

Example

f ( x, y )

Prepared by : M. Dokhi, Ph.D.

1
xy
16

Example

Prepared by : M. Dokhi, Ph.D.

Example

Prepared by : M. Dokhi, Ph.D.

Example

Prepared by : M. Dokhi, Ph.D.

Covariance: Example
20 of 45
A joint pdf of two random variables X, Y is given by
xy
96
f ( x, y )
0

for

0 x 4, 1 y 5
otherwise

8
xy
E[ X ] x dxdy
96
3
x 0 y 1
4

31
xy
y
dxdy

x0 y1 96
9
4

E[Y ]

E[ XY ]

xyf ( x, y)dxdy

1
xy
xy dxdy

96
96

x 0 y 1
0
4

2 2
x
y dydx
1

248
27

Thus, Cov(X,Y) = E[XY] E[X]E[Y] :

XY
9:01

248 8 31
E[ XY ] E[ X ]. E[Y ]
0
27 3 9

Prepared by : M. Dokhi, Ph.D.

Variance of a function
21 of 45

9:01

Some properties of variance

If c is a constant, Var[cX] = c2Var[X]

If X and Y are independent random variables, then

Var[X Y] = Var[X] + Var[Y]
Var[aX + bX]= a2Var[X] + b2Var[Y],
where a, b are constants

Prepared by : M. Dokhi, Ph.D.

Correlation Coefficient
22 of 45

9:01

Correlation is another measure of the strength of

dependence between two random variables.
It scales the covariance by the standard deviation of
each variable.
If X and Y are independent, then = 0, but = 0 does
not imply independence

Prepared by : M. Dokhi, Ph.D.

Correlation Coefficient: Problem

23 of 45

Assume the length X in minutes of a particular type of

telephone conversation is a random variable with
probability density function

1 x / 5
f (X ) e
5

0 x

Determine

The mean length E(X) of this telephone

conversation.

Find the variance and standard deviation of X

Find E[( X 5) 2 ]

9:01

Prepared by : M. Dokhi, Ph.D.

Correlation Coefficient: Problem

24 of 45
Use integration by part:

E[ X ]

9:01

Prepared by : M. Dokhi, Ph.D.

1
xe x 5 dx

50
uv vdu

1
5 xe x 5 5e x 5 dx
0
5

e x 5
1
x 5

5 xe 5
5
1 5 0

1
5 xe x 5 25e x 5 0
5
1
25
5
5

ux
du dx

dv e x 5 dx
v 5e x 5

Correlation Coefficient: Problem

25 of 45

E[ X ]
2

1 2 x 5
x e dx

50
uv vdu

1
2 x 5
x 5
5x e
10 xe dx
0
5
1
5 x 2 e x 5 10 5 xe x 5 5e x 5 dx
5

1
5 x 2 e x 5 10 5 xe x 5 25e x 5 0
5
1
5 x 2 e x 5 50 xe x 5 250e x 5
5
250
50
5

u x2
du 2 xdx

var( X ) E ( X 2 ) ( E ( X )) 2 50 5 2 25
9:01

std ( X ) var(X ) 5

Prepared by : M. Dokhi, Ph.D.

ux
du dx

dv e x 5 dx
v 5e x 5

dv e x 5 dx
v 5e x 5

Correlation Coefficient: problem

26 of 45

E[(X + 5)2]

9:01

Prepared by : M. Dokhi, Ph.D.

= E[(X2 + 10X + 25)]

= E[X2] + 10E[X] + E[25]
= 50 + 10[5] + 25
= 125

Correlation Coefficient: problem

27 of 45

The joint density function of X and Y is given by

1
200

f ( x, y )

20 x y 40

elsewhere

Find the covariance and correlation coefficient of X and Y

9:01

Prepared by : M. Dokhi, Ph.D.

Correlation Coefficient: problem

30 of 45

Consider the joint density function

16 y
f ( x, y ) 3
x

f ( x, y ) 0
Compute

9:01

Prepared by : M. Dokhi, Ph.D.

f(x),

x >2; 0 < y < 1;

elsewhere;

f(y), E[X],

E[Y],

E[XY],

XY, XY.

Moment
31 of 45

The kth moment about the origin of a random variable X is

x k f ( x)
if X is discrete

x
k
'k E[ X ]
k
x f ( x) if X is continuous

The kth moment about the mean is

k E[ X E ( X )] k E ( X ) k

9:01

Prepared by : M. Dokhi, Ph.D.

Moment
32 of 45

Moments are useful in characterizing some features of the

distribution
The first and the second moment about the origin are given by

'1 E[ X ]

We can write the mean and variance of a random variable as

'
1

' 2 E[ X 2 ]
2 ' 2 ( '1 ) 2

The second moment about the mean is the variance.

2 E[ X ] 2 2

9:01

The third moment about the mean is a measure of skewness of

a distribution.

Prepared by : M. Dokhi, Ph.D.

Moment Generating Function (MGF)

33 of 45

9:01

Moment-generating function is used to determine the moments of

distribution
It will exist only if the sum or integral converges.
If a moment-generating function of X does exist, it can be used to
generate all the moments of that variable.

Prepared by : M. Dokhi, Ph.D.

Moment Generating Function (MGF)

34 of 45

M X (t ) e txi f ( xi )
i 1

M X' (t ) xi e txi f ( xi )
i 1

9:01

Prepared by : M. Dokhi, Ph.D.

(r )
X

(t ) x ir e txi f ( xi )
i 1

Moment Generating Function (MGF)

35 of 45

(r )
X

(t 0) x i f ( xi )
r

i 1

E[ X ], E[ X ],..., E[ X ]

9:01

Prepared by : M. Dokhi, Ph.D.

Moment Generating Function (MGF)

36 of 45

Find the moment-generating function of the binomial random

variable X and then use it to verify that np and 2 npq
n
tx n x n x
M X (t ) e p q
x 0
x
n
The last sum is
n t x n x
the binomial
pe q
expansion of
x 0 x
(pet+q)n

pe q

First derivation, E[X]

Second derivation,

dM X (t )
n pet q
dt

E[X2]

9:01

Prepared by : M. Dokhi, Ph.D.

n 1

pet

d 2 M X (t )
t
t

np
e
n

1
pe
q
2
dt

1
'
Setting t = 0 we get E[ X ] 1 np

Therefore,

1' np

n2

pe t pe t q

E[ X 2 ] 2' npn 1 p 1

2 2' 2 np(1 p) npq

n 1

et

Conditional Expectation
37 of 45

If X and Y are jointly distributed R.V. then the conditional

expectation of Y given X=x is given by

E(Y | x) y f ( y | x)

X dan Y diskrit

E (Y | x) y f ( y | x)dy
y

9:01

Prepared by : M. Dokhi, Ph.D.

X dan Y kontinu

Conditional Expectation: latihan

38 of 45

f ( x, y) x y 0 x 1,0 y 1
Carilah

E ( y | x)

f ( y | x)

9:01

f ( x, y )
x y

f ( x)
x 0.5

Prepared by : M. Dokhi, Ph.D.

0 y 1

Conditional Expectation: latihan

39 of 45
Jika X dan Y memiliki joint distribusi maka

E X EY Y | X EY (Y )

E X EY Y | X E (Y | x). f ( x)dx
x

y f ( y | x ) f ( x )dydx
x y

M
EY (Y )
9:01

Prepared by : M. Dokhi, Ph.D.

Joint MGF
40 of 45
Definisi : Joint MGF dari vector random X ( X 1 , X 2 ,..., X k ) adalah

M X (t ) E exp ti X i

i 1
Dimana

9:01

t (t 1,..., t k )

Prepared by : M. Dokhi, Ph.D.

dan

h ti h

for h > 0