Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
0% found this document useful (0 votes)
342 views

Bivariate Distributions

The document discusses bivariate distributions, which describe the joint probability distribution of two random variables. It defines the joint distribution function and its properties, including that it must be non-negative and converge to 0 and 1 as its arguments approach negative and positive infinity respectively. It also defines marginal distribution functions, joint discrete density functions, conditional distributions, expectation, covariance, correlation, and conditional expectation. An example is provided to illustrate these concepts.

Uploaded by

Aiman imam
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
342 views

Bivariate Distributions

The document discusses bivariate distributions, which describe the joint probability distribution of two random variables. It defines the joint distribution function and its properties, including that it must be non-negative and converge to 0 and 1 as its arguments approach negative and positive infinity respectively. It also defines marginal distribution functions, joint discrete density functions, conditional distributions, expectation, covariance, correlation, and conditional expectation. An example is provided to illustrate these concepts.

Uploaded by

Aiman imam
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Bivariate Distributions

Joint Distribution function: Let X and Y be two random variables defined on the same probability space
(𝛺, 𝐴̃, 𝑃[. ]). Then (X, Y) is called a two-dimensional random variable. The joint cumulative distribution
function or joint distribution function of X and Y, denoted by 𝐹𝑋,𝑌 (𝑥, 𝑦), is defined as

𝐹𝑋,𝑌 (𝑥, 𝑦) = 𝑃[𝑋 ≤ 𝑥, 𝑌 ≤ 𝑦] ∀ 𝑥, 𝑦 ∈ 𝑅.

It may be observed that the joint distribution function is a function of two variables and its domain is the
xy-plane. Sometimes we write 𝐹𝑋,𝑌 (𝑥, 𝑦) 𝑎𝑠 𝐹(𝑥, 𝑦).

Properties of Joint Distribution function:


1. If 𝑥1 < 𝑥2 𝑎𝑛𝑑 𝑦1 < 𝑦2 , then (rectangle rule)

𝑃[𝑥1 < 𝑋 ≤ 𝑥2 , 𝑦1 < 𝑌 ≤ 𝑦2 ] = 𝐹(𝑥2 , 𝑦2 ) − 𝐹(𝑥2 , 𝑦1 ) − 𝐹(𝑥1 , 𝑦2 ) + 𝐹(𝑥1 , 𝑦1 ) ≥ 0

2. (a). 𝐹(−∞, 𝑦) = lim 𝐹(𝑥, 𝑦) = 0 ∀ 𝑦 ∈ 𝑅.


𝑥→−∞

(b). 𝐹(𝑥, −∞) = lim 𝐹(𝑥, 𝑦) = 0 ∀ 𝑥 ∈ 𝑅.


𝑦→−∞

(c). 𝐹(𝑥, −∞) = lim . 𝐹(𝑥, 𝑦) = 1.


𝑥→∞
𝑦 →∞

3. 𝐹(𝑥, 𝑦) is right continuous in each argument i.e.

lim 𝐹(𝑥 + ℎ, 𝑦) = lim 𝐹(𝑥, 𝑦 + ℎ) = 𝐹(𝑥, 𝑦).


ℎ→0+ ℎ→0+

Remark: Any function of two variables which fails to satisfy one of the above three conditions is not a

joint distribution function.

𝑒 −(𝑥+𝑦) , 𝑥 ≥ 0, 𝑦 ≥ 0
Example 1. Show that the bivariate function: 𝐹(𝑥, 𝑦) = { is not a joint
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒,

distribution function.

Solution: lim . 𝐹(𝑥, 𝑦) = lim . 𝑒−(𝑥+𝑦) = lim 𝑒−𝑥 . lim 𝑒−𝑦 = 0 ≠ 1


𝑥→∞ 𝑥→∞ 𝑥→∞ 𝑦→∞
𝑦 →∞ 𝑦 →∞

Hence, 𝐹(𝑥, 𝑦) is not a joint distribution function.

Marginal cumulative distribution function: If 𝐹𝑋,𝑌 (𝑥, 𝑦) is the joint cumulative distribution function of

two random variables X and Y, then 𝐹𝑋 (𝑥, 𝑦) 𝑎𝑛𝑑 𝐹𝑌 (𝑥, 𝑦) are called marginal distribution functions of X

and Y respectively, are defined as

𝐹𝑋 (𝑥) = 𝑃[𝑋 ≤ 𝑥] = 𝑃[𝑋 ≤ 𝑥, 𝑌 < ∞] = lim 𝐹𝑋,𝑌 (𝑥, 𝑦) = 𝐹𝑋,𝑌 (𝑥, ∞).
𝑦→∞
𝐹𝑌 (𝑦) = 𝑃[𝑌 ≤ 𝑦] = 𝑃[𝑋 < ∞, 𝑌 ≤ 𝑦] = lim 𝐹𝑋,𝑌 (𝑥, 𝑦) = 𝐹𝑋,𝑌 (∞, 𝑦).
𝑥→∞

Remark: 𝐹𝑋 (−∞) = 0, 𝐹𝑋 (∞) = 1, 𝐹𝑌 (−∞) = 0 𝑎𝑛𝑑 𝐹𝑌 (∞) = 1.

Joint Discrete Density function: If X and Y are jointly discrete random variables, then the joint discrete

density function of X and Y, defined as

𝑓𝑋,𝑌 (𝑥, 𝑦) = 𝑃[𝑋 = 𝑥, 𝑌 = 𝑦] = 𝑃[𝑋 = 𝑥 ∩ 𝑌 = 𝑦].

 𝑓𝑋 (𝑥) 𝑎𝑛𝑑 𝑓𝑌 (𝑦) are called marginal density functions of X and Y respectively, are defined as

𝑓𝑋 (𝑥) = 𝑃[𝑋 = 𝑥] 𝑎𝑛𝑑 𝑓𝑌 (𝑦) = 𝑃[𝑌 = 𝑦].

 Conditional discrete density function of Y given 𝑋 = 𝑥, denoted by 𝑓𝑌|𝑋 (𝑦|𝑥), is defined as

𝑓𝑋,𝑌 (𝑥,𝑦) 𝑃[𝑋=𝑥,𝑌=𝑦]


𝑓𝑌|𝑋 (𝑦|𝑥) = 𝑓𝑋 (𝑥)
= 𝑃[𝑋=𝑥]
, 𝑖𝑓 𝑓𝑋 (𝑥) > 0.

 Conditional discrete density function of X given 𝑌 = 𝑦, denoted by 𝑓𝑋|𝑌 (𝑥|𝑦), is defined as

𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑃[𝑋 = 𝑥, 𝑌 = 𝑦]


𝑓𝑋|𝑌 (𝑥|𝑦) = = , 𝑖𝑓 𝑓𝑌 (𝑦) > 0
𝑓𝑌 (𝑦) 𝑃[𝑌 = 𝑦]

 If X and Y are independent discrete random variable, then

𝑓𝑋,𝑌 (𝑥, 𝑦) = 𝑃[𝑋 = 𝑥, 𝑌 = 𝑦] = 𝑓𝑋 (𝑥). 𝑓𝑌 (𝑦) = 𝑃[𝑋 = 𝑥]. 𝑃[𝑌 = 𝑦] ∀ (𝑥, 𝑦) 𝑜𝑓 (𝑋, 𝑌)

The joint probability function of X and Y is usually represented in the form of the following table:

Y 𝒚𝟏 𝒚𝟐 … 𝒚𝒋 … 𝒚𝒎 𝒇𝑿 (𝒙)
X
𝒙𝟏 𝑝11 𝑝12 … 𝑝1𝑗 … 𝑝1𝑚 𝑓𝑋 (𝑥1 )

𝒙𝟐 𝑝21 𝑝22 … 𝑝2𝑗 … 𝑝2𝑚 𝑓𝑋 (𝑥2 )


𝒙𝒊 𝑝𝑖1 𝑝𝑖2 … 𝑝𝑖𝑗 … 𝑝𝑖𝑚 𝑓𝑋 (𝑥𝑖 )


𝒙𝒏 𝑝𝑛1 𝑝𝑛2 … 𝑝𝑛𝑗 … 𝑝𝑛𝑚 𝑓𝑋 (𝑥𝑛 )

𝒇𝒀 (𝒚) 𝑓𝑌 (𝑦1 ) 𝑓𝑌 (𝑦2 ) … 𝑓𝑌 (𝑦𝑗 ) … 𝑓𝑌 (𝑦𝑚 ) 1

From the following table we conclude that:


i. 𝑝𝑖𝑗 = 𝑓𝑋,𝑌 (𝑥𝑖 , 𝑦𝑗 ) = 𝑃[𝑋 = 𝑥𝑖 , 𝑌 = 𝑦𝑗 ] = 𝑃[𝑋 = 𝑥𝑖 ∩ 𝑌 = 𝑦𝑗 ] is the joint discrete density function

of X and Y.

ii. 𝑓𝑋 (𝑥𝑖 ) = 𝑃[𝑋 = 𝑥𝑖 ] = ∑𝑚


𝑗=1 𝑝𝑖𝑗 is the marginal discrete density function of X.

iii. 𝑓𝑌 (𝑦𝑗 ) = 𝑃[𝑌 = 𝑦𝑗 ] = ∑𝑛𝑖=1 𝑝𝑖𝑗 is the marginal discrete density function of Y.

iv. ∑𝑚 𝑛 𝑛 𝑚
𝑗=1 𝑓𝑌 (𝑦𝑗 ) = ∑𝑖=1 𝑓𝑋 (𝑥𝑖 ) = ∑𝑖=1 ∑𝑗=1 𝑝𝑖𝑗 = 1.

Expectation: Let (X, Y) be a two dimensional discrete random variable with joint discrete density

𝑓𝑋,𝑌 (𝑥, 𝑦). The expectation or expected value of a function 𝑔(𝑋, 𝑌), denoted as 𝐸[𝑔(𝑋, 𝑌)], is defined as

𝐸[𝑔(𝑋, 𝑌)] = ∑𝑥 ∑𝑦 𝑔(𝑥, 𝑦). 𝑓𝑋,𝑌 (𝑥, 𝑦).

## In particular, 𝐸[𝑋𝑌] = ∑𝑥 ∑𝑦 𝑥𝑦. 𝑓𝑋,𝑌 (𝑥, 𝑦), 𝐸[𝑋] = ∑𝑥 𝑥. 𝑓𝑋 (𝑥) and 𝐸[𝑌] = ∑𝑦 𝑦. 𝑓𝑌 (𝑦).

Covariance: Let X and Y be two random variables defined on the same probability space. The

covariance of X and Y, denoted as 𝑐𝑜𝑣(𝑋, 𝑌), is defined as

𝑐𝑜𝑣(𝑋, 𝑌) = 𝐸[{𝑋 − 𝐸[𝑋]}{𝑌 − 𝐸[𝑌]}] OR

𝑐𝑜𝑣(𝑋, 𝑌) = 𝐸[𝑋𝑌] − 𝐸[𝑋]. 𝐸[𝑌]

Correlation coefficient: The correlation coefficient of two random variables X and Y, denoted by 𝜌(𝑋, 𝑌),

𝑐𝑜𝑣(𝑋,𝑌) 𝑐𝑜𝑣(𝑋,𝑌)
is defined as 𝜌(𝑋, 𝑌) = = , where 𝜎𝑋 > 0 and 𝜎𝑌 > 0.
√𝑣𝑎𝑟[𝑋].𝑣𝑎𝑟[𝑌] √𝜎𝑋 .𝜎𝑌

# It can be verify that −1 ≤ 𝜌(𝑋, 𝑌) ≤ 1.

## Two random variables X and Y are uncorrelated if

𝜌(𝑋, 𝑌) = 0 𝑜𝑟 𝑐𝑜𝑣(𝑋, 𝑌) = 0

⟹ X and Y are independent random variables.

Conditional Expectation: Let (𝑋, 𝑌) be a two dimensional discrete random variable with joint discrete

density 𝑓𝑋,𝑌 (𝑥, 𝑦), then the conditional expectation of 𝑔(𝑋, 𝑌) given 𝑋 = 𝑥, denoted by 𝐸[𝑔(𝑋, 𝑌)|𝑋 = 𝑥],

is defined as 𝐸[𝑔(𝑋, 𝑌)|𝑋 = 𝑥] = ∑𝑗 𝑔(𝑥, 𝑦𝑗 ). 𝑓𝑌|𝑋 (𝑦𝑗 |𝑥).

In particular, 𝐸[𝑌|𝑋 = 𝑥] = ∑𝑗 𝑦𝑗 . 𝑓𝑌|𝑋 (𝑦𝑗 |𝑥) = ∑𝑗 𝑦𝑗 . 𝑃[𝑌 = 𝑦𝑗 |𝑋 = 𝑥].


Example 1.

X and Y have joint p.d.f.: Y −𝟏 𝟎 1


X
Find (i) marginal distribution of X and Y, 𝟎 𝑎 2𝑎 a
𝟏 3𝑎 2𝑎 a
(ii) Conditional distribution of X given Y = 2.
2 2a a 2a
1
Solution: As 𝑓𝑋,𝑌 (𝑥, 𝑦) is a p.d.f. ⟹ 15𝑎 = 1 ⟹ 𝑎 = 15
.

X −𝟏 𝟎 1 𝑓𝑌 (𝑦) (i) Marginal distribution of X. From the table, we get,


Y
𝟎 𝑎 2𝑎 a 4a 𝑓𝑋 (−1) = 6𝑎, 𝑓𝑋 (0) = 5𝑎 𝑎𝑛𝑑 𝑓𝑋 (1) = 4𝑎.
𝟏 3𝑎 2𝑎 a 6a
Marginal distribution of X. From the table, we get,
2 2a a 2a 5a
𝑓𝑋 (𝑥) 6a 5a 4a 15a 𝑓𝑌 (0) = 4𝑎, 𝑓𝑌 (1) = 6𝑎 𝑎𝑛𝑑 𝑓𝑌 (2) = 5𝑎.

(ii) Conditional distribution of X given Y = 2 is obtained as

𝑃[𝑋=𝑥∩𝑌=2]
𝑓𝑋|𝑌 (𝑋 = 𝑥|𝑦 = 2) = 𝑃[𝑌=2]
.

𝑃[𝑋=−1∩𝑌=2] 2𝑎 2
Therefore, 𝑓𝑋|𝑌 (𝑋 = −1|𝑦 = 2) = 𝑃[𝑌=2]
= 5𝑎 = 5.

𝑃[𝑋=0∩𝑌=2] 𝑎 1
𝑓𝑋|𝑌 (𝑋 = 0|𝑦 = 2) = = = .
𝑃[𝑌=2] 5𝑎 5

𝑃[𝑋=1∩𝑌=2] 2𝑎 2
𝑓𝑋|𝑌 (𝑋 = 1|𝑦 = 2) = 𝑃[𝑌=2]
= 5𝑎 = 5.

Example 2. Two tetrahedral with sides numbered 1 to 4 are tossed. Let X denote the number on the

downturned face of the first tetrahedron and Y the larger of the downturned numbers. Find

i. The joint density function of X and Y.

ii. The marginal density functions of X and Y.

iii. 𝑃[𝑋 ≤ 2, 𝑌 ≤ 3].

iv. The conditional distribution of Y given X = 2 and X = 3.

v. 𝐸[𝑌|𝑋 = 2] 𝑎𝑛𝑑 𝐸[𝑌|𝑋 = 3].

vi. 𝜌(𝑋, 𝑌).

Solution: (i) From the two tetrahedrons, total number of outputs (First face number, second face number)

are as follows:
(1, 1), (1, 2), (1, 3), (1, 4), (2, 1), (2, 2), (2, 3), (2, 4), (3, 1), (3, 2), (3, 3), (3, 4), (4, 1), (4, 2), (4, 3), (4, 4).

It is given that X denotes the number on the downturned face on the first tetrahedron and Y the larger of

the numbers on the two downturned faces. Then (X, Y) takes the values:

(1, 1), (1, 2), (1, 3), (1, 4), (2, 2), (2, 3), (2, 4), (3, 3), (3, 4), (4, 4).

The joint discrete density function of X and Y is given below:

(x, y) (1, 1) (1, 2) (1, 3) (1, 4) (2, 2) (2, 3) (2, 4) (3, 3) (3, 4) (4, 4)

𝑓𝑋,𝑌 (𝑥, 𝑦) 1/16 1/16 1/16 1/16 2/16 1/16 1/16 3/16 1/16 4/16

We can also write the above data as follows:

Y 1 2 3 4 𝒇𝑿 (𝒙)
X
1 1 1 1 4
1
16 16 16 16 16

2 1 1 4
2 0
16 16 16 16

3 1 4
3 0 0
16 16 16

4 4
4 0 0 0
16 16

1 3 5 7
𝒇𝒀 (𝒚) 1
16 16 16 16

(ii) Marginal distribution of X, From the above table, we get


4 4 4 4
𝑓𝑋 (1) = 16 , 𝑓𝑋 (2) = 16 , 𝑓𝑋 (3) = 16 𝑎𝑛𝑑 𝑓𝑋 (4) = 16 .

(iii) Marginal distribution of Y, From the above table, we get

1 3 5 7
𝑓𝑌 (1) = 16 , 𝑓𝑌 (2) = 16 , 𝑓𝑌 (3) = 16 𝑎𝑛𝑑 𝑓𝑌 (4) = 16 .

(iv) 𝑃[𝑋 ≤ 2, 𝑌 ≤ 3] = 𝑃[𝑋 = 1, 𝑌 = 1] + 𝑃[𝑋 = 1, 𝑌 = 2] + 𝑃[𝑋 = 1, 𝑌 = 3]

+𝑃[𝑋 = 2, 𝑌 = 1] + 𝑃[𝑋 = 2, 𝑌 = 2] + 𝑃[𝑋 = 2, 𝑌 = 3]


1 1 1 2 1 3
𝑃[𝑋 ≤ 2, 𝑌 ≤ 3] = 16 + 16 + 16 + 0 + 16 + 16 = 8 .

(v) The conditional distribution of Y given X = 2, we have


𝑃[𝑌=𝑦 ∩ 𝑋=2]
𝑃[𝑌 = 𝑦|𝑋 = 2] = 𝑃[𝑋=2]
.

𝑃[𝑌=1 ∩ 𝑋=2] 0
Now 𝑃[𝑌 = 1|𝑋 = 2] = 𝑃[𝑋=2]
= 4 = 0.
16

2
𝑃[𝑌=2 ∩ 𝑋=2] 16 1
𝑃[𝑌 = 2|𝑋 = 2] = 𝑃[𝑋=2]
= 4 =2.
16

1
𝑃[𝑌=3 ∩ 𝑋=2] 16 1
𝑃[𝑌 = 3|𝑋 = 2] = 𝑃[𝑋=2]
= 4 =4.
16

1
𝑃[𝑌=4 ∩ 𝑋=2] 16 1
𝑃[𝑌 = 4|𝑋 = 2] = = 4 = .
𝑃[𝑋=2] 4
16

The conditional distribution of Y given X = 3, we have

𝑃[𝑌=𝑦 ∩ 𝑋=3]
𝑃[𝑌 = 𝑦|𝑋 = 3] = .
𝑃[𝑋=3]

𝑃[𝑌=1 ∩ 𝑋=3] 0
Now 𝑃[𝑌 = 1|𝑋 = 3] = 𝑃[𝑋=3]
= 4 = 0.
16

𝑃[𝑌=2 ∩ 𝑋=3] 0
𝑃[𝑌 = 2|𝑋 = 3] = 𝑃[𝑋=3]
= 4 = 0.
16

3
𝑃[𝑌=3 ∩ 𝑋=3] 16 3
𝑃[𝑌 = 3|𝑋 = 3] = 𝑃[𝑋=3]
= 4 =4.
16

1
𝑃[𝑌=4 ∩ 𝑋=3] 16 1
𝑃[𝑌 = 4|𝑋 = 3] = 𝑃[𝑋=3]
= 4 =4.
16

(vi) Now conditional expectation 𝐸[𝑌|𝑋 = 2] = ∑𝑗 𝑦𝑗 . 𝑃[𝑌 = 𝑦𝑗 |𝑋 = 2]

= 1. 𝑃[𝑌 = 1|𝑋 = 2] + 2. 𝑃[𝑌 = 2|𝑋 = 2] + 3. 𝑃[𝑌 = 3|𝑋 = 2] + 4. 𝑃[𝑌 = 4|𝑋 = 2]


1 1 1 11
= 1.0 + 2. + 3. + 4. = .
2 4 4 4

Again conditional expectation 𝐸[𝑌|𝑋 = 3] = ∑𝑗 𝑦𝑗 . 𝑃[𝑌 = 𝑦𝑗 |𝑋 = 3]

= 1. 𝑃[𝑌 = 1|𝑋 = 3] + 2. 𝑃[𝑌 = 2|𝑋 = 3] + 3. 𝑃[𝑌 = 3|𝑋 = 3] + 4. 𝑃[𝑌 = 4|𝑋 = 3]


3 1 13
= 1.0 + 2.0 + 3. + 4. = .
4 4 4

(vii) Now we have to find 𝜌(𝑋, 𝑌) i.e. the correlation coefficient between X and Y. From the table, we have

4 4 4 4 5
𝐸[𝑋] = ∑𝑥 𝑥. 𝑓𝑋 (𝑥) = 1 × 16 + 2 × 16 + 3 × 16 + 4 × 16 = 2 .

4 4 4 4 15
𝐸[𝑋 2 ] = ∑𝑥 𝑥 2 . 𝑓𝑋 (𝑥) = 1 × 16 + 4 × 16 + 9 × 16 + 16 × 16 = 2
.
1 3 5 7 25
𝐸[𝑌] = ∑𝑦 𝑦. 𝑓𝑌 (𝑦) = 1 × 16 + 2 × 16 + 3 × 16 + 4 × 16 = 8
.

1 3 5 7 85
𝐸[𝑌 2 ] = ∑𝑦 𝑦 2 . 𝑓𝑌 (𝑦) = 1 × 16 + 4 × 16 + 9 × 16 + 16 × 16 = 8
.

𝐸[𝑋𝑌] = ∑ ∑ 𝑥𝑦. 𝑓𝑋,𝑌 (𝑥, 𝑦)


𝑥 𝑦

1 1 1 1 2 1 1
= (1 × 16 + 2 × 16 + 3 × 16 + 4 × 16) + (2 × 0 + 4 × 16 + 6 × 16 + 8 × 16).

3 1 4
+ (3 × 0 + 6 × 0 + 9 × 16 + 12 × 16) + (4 × 0 + 8 × 0 + 12 × 0 + 16 × 16).

135
𝐸[𝑋𝑌] = 16
.

15 5 2 5
𝑣𝑎𝑟[𝑋] = 𝐸[𝑋 2 ] − (𝐸[𝑋])2 = 2
− (2
) =4.

85 25 2 55
𝑣𝑎𝑟[𝑌] = 𝐸[𝑌 2 ] − (𝐸[𝑌])2 = 8
− ( 8
) = 64 .

135 5 25 5
𝑐𝑜𝑣(𝑋, 𝑌) = 𝐸[𝑋𝑌] − 𝐸[𝑋]. 𝐸[𝑌] = − . = .
16 2 8 8

5
𝑐𝑜𝑣(𝑋,𝑌) 8 2
Hence 𝜌(𝑋, 𝑌) = = =
√𝑣𝑎𝑟[𝑋].𝑣𝑎𝑟[𝑌] 5 55
√ . √11
4 64

Exercises:

1. For the following bivariate probability distribution of X and Y:

𝑌 1 2 3 4 5 6 Find: (i) 𝑃[𝑋 ≤ 1, 𝑌 = 2], (ii) 𝑃[𝑋 ≤ 1]


𝑋
0 0 0 1/32 2/32 2/32 3/32 (iii) 𝑃[𝑌 = 3], (iv) 𝑃[ 𝑌 ≤ 3] and
1 1/16 1/16 1/8 1/8 1/8 1/8
2 1/32 1/32 1/64 1/64 0 2/64 (v) 𝑃[𝑋 < 3, 𝑌 ≤ 4]

Ans. 1/16, 7/8, 11/64, 23/64 and 9/16.

2. Consider two tetrahedral with sides numbered 1 to 4. Let X denote the smaller of the two downturned

numbers and Y the larger. (i) Find the joint density function of X and Y, (ii) find 𝑃[𝑋 ≥ 2, 𝑌 ≥ 2], (iii)

Find the mean and variance of X and Y, (iv) Find the conditional distribution of Y given X for each of

the possible values of X, and (v) Find the correlation coefficient of X and Y.

Ans. (i) Hint: 𝑋 ≤ 𝑌 always (ii) 9/16 (iii) 55/64 (iv) 0, 1/5, 2/5, 2/5, 0, 0, 0, 1 and other self (v) 5/11.
3. Three fair coins are tossed. Let X denote the number of heads on the first two coins, and let Y denote

the numbers of tails on the last two coins. (i) Find the joint distribution of X and Y. (ii) Find the

conditional distribution of Y given X = 1. (iii) Find cov (X, Y), and (iv) Find correlation coefficient of

X and Y. Ans. (i) self (ii) ¼, ½, ¼ (iii) -1/4 (iv) -0.5

Joint Continuous Density function: A two dimensional random variable (X, Y) is said to be continuous

if and only if there exists a function 𝑓𝑋,𝑌 (𝑥, 𝑦) ≥ 0 𝑜𝑟 𝑠𝑖𝑚𝑝𝑙𝑦 𝑓(𝑥, 𝑦) ≥ 0 such that
𝑥 𝑦
𝐹𝑋,𝑌 (𝑥, 𝑦) = ∫−∞ ∫−∞ 𝑓𝑋,𝑌 (𝑢, 𝑣) 𝑑𝑢𝑑𝑣 for all (𝑥, 𝑦) ∈ 𝑅 2

The function 𝐹𝑋,𝑌 (𝑥, 𝑦) = 𝑃[𝑋 ≤ 𝑥, 𝑌 ≤ 𝑦] is joint distribution function of X and Y. The function

𝑓𝑋,𝑌 (𝑥, 𝑦) is joint probability density function of X and Y which satisfies the following properties:

i. 𝑓𝑋,𝑌 (𝑥, 𝑦) ≥ 0 𝑓𝑜𝑟 𝑎𝑙𝑙 (𝑥, 𝑦) ∈ 𝑅 2


∞ ∞
ii. ∫−∞ ∫−∞ 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = 1.

 𝑓𝑋 (𝑥) 𝑎𝑛𝑑 𝑓𝑌 (𝑦) are called marginal density functions of X and Y respectively, are defined as
∞ ∞
𝑓𝑋 (𝑥) = ∫−∞ 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑦 𝑎𝑛𝑑 𝑓𝑌 (𝑦) = ∫−∞ 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥.

 Conditional continuous density function of Y given 𝑋 = 𝑥, denoted by 𝑓𝑌|𝑋 (𝑦|𝑥), is defined as

𝑓𝑋,𝑌 (𝑥,𝑦)
𝑓𝑌|𝑋 (𝑦|𝑥) = 𝑓𝑋 (𝑥)
, 𝑖𝑓 𝑓𝑋 (𝑥) > 0.

 Conditional discrete density function of X given 𝑌 = 𝑦, denoted by 𝑓𝑋|𝑌 (𝑥|𝑦), is defined as

𝑓𝑋,𝑌 (𝑥, 𝑦)
𝑓𝑋|𝑌 (𝑥|𝑦) = , 𝑖𝑓 𝑓𝑌 (𝑦) > 0
𝑓𝑌 (𝑦)

 If X and Y are independent continuous random variables, then

𝑓𝑋,𝑌 (𝑥, 𝑦) = 𝑓𝑋 (𝑥). 𝑓𝑌 (𝑦) ∀ (𝑥, 𝑦) 𝑜𝑓 (𝑋, 𝑌)

Expectation: Let (X, Y) be a two dimensional continuous random variables with joint probability density

function 𝑓𝑋,𝑌 (𝑥, 𝑦). The expectation or expected value of a function 𝑔(𝑋, 𝑌), denoted as 𝐸[𝑔(𝑋, 𝑌)], is

defined as
∞ ∞
𝐸[𝑔(𝑋, 𝑌)] = ∫−∞ ∫−∞ 𝑔(𝑥, 𝑦). 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥𝑑𝑦.
∞ ∞ ∞ ∞
## In particular, 𝐸[𝑋𝑌] = ∫−∞ ∫−∞ 𝑥𝑦. 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥𝑑𝑦, 𝐸[𝑋] = ∫−∞ ∫−∞ 𝑥. 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥𝑑𝑦 and

∞ ∞
𝐸[𝑌] = ∫−∞ ∫−∞ 𝑦. 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥𝑑𝑦.

Conditional Expectation: Let (𝑋, 𝑌) be a two dimensional continuous random variable with joint

continuous density function 𝑓𝑋,𝑌 (𝑥, 𝑦), then the conditional expectation of 𝑔(𝑋, 𝑌) given 𝑋 = 𝑥, denoted

by 𝐸[𝑔(𝑋, 𝑌)|𝑋 = 𝑥], is defined as 𝐸[𝑔(𝑋, 𝑌)|𝑋 = 𝑥] = ∫−∞ 𝑔(𝑥, 𝑦). 𝑓𝑌|𝑋 (𝑦|𝑥) 𝑑𝑦.


In particular, 𝐸[𝑌|𝑋 = 𝑥] = ∫−∞ 𝑦. 𝑓𝑌|𝑋 (𝑦|𝑥) 𝑑𝑦.

Regression curve:

1. The relation 𝑦 = 𝐸[𝑌|𝑋 = 𝑥] is called the regression curve of Y on 𝑋 = 𝑥.

2. The relation 𝑥 = 𝐸[𝑋|𝑌 = 𝑦] is called the regression curve of X on 𝑌 = 𝑦.

Example 1. Find 𝑘 so that 𝑓(𝑥, 𝑦) = 𝑘𝑥𝑦, 1 ≤ 𝑥 ≤ 𝑦 ≤ 2 will be a joint probability density function.

∞ ∞ 2 2 2 1
Solution: We have 1 = ∫−∞ ∫−∞ 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫1 ∫𝑥 𝑘𝑥𝑦 𝑑𝑦𝑑𝑥 = ∫1 𝑥. 2 (4 − 𝑥 2 ) 𝑑𝑥 = 9𝑘/8

⟹ 𝑘 = 8/9.

(𝑥 + 𝑦), 0 ≤ 𝑥 ≤ 1, 0 ≤ 𝑦 ≤ 1
Example 2. Let the joint p.d.f. of X and Y be 𝑓(𝑥, 𝑦) = {
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒.
1 1
Find (i) 𝑃 [0 < 𝑋 < 2 , 0 < 𝑌 < 4], (ii) 𝐸[𝑋], 𝐸[𝑌], 𝐸[𝑋𝑌] 𝑎𝑛𝑑 𝐸[𝑋 + 𝑌] (iii) 𝜎(𝑋, 𝑌).

1 1
1 1
Solution: We have 𝑃 [0 < 𝑋 < 2 , 0 < 𝑌 < 4] = ∫02 [∫04(𝑥 + 𝑦) 𝑑𝑦] 𝑑𝑥

1/2 1 1/4 1/2 1 1 3


= ∫0 |𝑥𝑦 + 2 𝑦 2 | 𝑑𝑥 = ∫0 (4 𝑥 + 32) 𝑑𝑥 = 64 .
0

∞ ∞ 1 1 1 1 7
Now 𝐸[𝑋] = ∫−∞ ∫−∞ 𝑥. 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫0 ∫0 𝑥. (𝑥 + 𝑦) 𝑑𝑥𝑑𝑦 = ∫0 (𝑥 2 + 2 𝑥) 𝑑𝑥 = 12 .

∞ ∞ 1 1 7
Similarly, 𝐸[𝑌] = ∫−∞ ∫−∞ 𝑦. 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫0 ∫0 𝑦. (𝑥 + 𝑦) 𝑑𝑥𝑑𝑦 = 12 .

∞ ∞ 1 1 1
Now 𝐸[𝑋𝑌] = ∫−∞ ∫−∞ 𝑥𝑦. 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫0 ∫0 𝑥𝑦. (𝑥 + 𝑦) 𝑑𝑥𝑑𝑦 = .
3

∞ ∞ 1 1
7
𝐸[𝑋 + 𝑌] = ∫ ∫ (𝑥 + 𝑦). 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫ ∫(𝑥 + 𝑦). (𝑥 + 𝑦) 𝑑𝑥𝑑𝑦 = .
6
−∞ −∞ 0 0

1
We have 𝑐𝑜𝑣[𝑋, 𝑌] = 𝐸[𝑋𝑌] − 𝐸[𝑋]. 𝐸[𝑌] = − 144 .
∞ ∞ 1 1
2]
5
𝐸[𝑋 = ∫ ∫ 𝑥 . 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫ ∫ 𝑥 2 . (𝑥 + 𝑦) 𝑑𝑥𝑑𝑦 =
2
.
12
−∞ −∞ 0 0

∞ ∞ 1 1
5
𝐸[𝑌 2 ] = ∫ ∫ 𝑦 2 . 𝑓(𝑥, 𝑦) 𝑑𝑥𝑑𝑦 = ∫ ∫ 𝑦 2 . (𝑥 + 𝑦) 𝑑𝑥𝑑𝑦 = .
12
−∞ −∞ 0 0

5 7 2 11
Now 𝑣𝑎𝑟[𝑋] = 𝐸[𝑋 2 ] − (𝐸[𝑋])2 = 12 − (12) = 144 .

11
𝑣𝑎𝑟[𝑌] = 𝐸[𝑌 2 ] − (𝐸[𝑌])2 = 144 .

1
𝑐𝑜𝑣(𝑋,𝑌) − 1
144
Hence 𝜌(𝑋, 𝑌) = = =− .
√𝑣𝑎𝑟[𝑋].𝑣𝑎𝑟[𝑌] 11 11 11
√ .
144 144

8𝑥𝑦, 𝑓𝑜𝑟 0 < 𝑥 < 𝑦 < 1


Example 3. Let the joint density function of X and Y be given by 𝑓(𝑥, 𝑦) = {
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒,

Find (i) 𝐸[𝑌|𝑋 = 𝑥], (ii) 𝐸[𝑋𝑌|𝑋 = 𝑥], (iii) 𝑣𝑎𝑟[𝑌|𝑋 = 𝑥] and (iv) regression curve of Y on 𝑋.

Solution: The marginal distribution of X and Y are given by


∞ 1

𝑓𝑋 (𝑥) = ∫ 𝑓(𝑥, 𝑦) 𝑑𝑦 = ∫ 8𝑥𝑦 𝑑𝑦 = 4𝑥(1 − 𝑥 2 ), 0 < 𝑥 < 1,


−∞ 𝑥

∞ 𝑦
𝑎𝑛𝑑 𝑓𝑌 (𝑦) = ∫−∞ 𝑓𝑋,𝑌 (𝑥, 𝑦) 𝑑𝑥 = ∫0 8𝑥𝑦 𝑑𝑥 = 4𝑦 3 , 0 < 𝑦 < 1.

We note that 𝑓(𝑥, 𝑦) ≠ 𝑓𝑋 (𝑥). 𝑓𝑌 (𝑦). Thus X and Y are not independent.

The conditional distribution of Y given X = x is given by

𝑓𝑋,𝑌 (𝑥,𝑦) 8𝑥𝑦 2𝑦


𝑓𝑌|𝑋 (𝑦|𝑥) = 𝑓𝑋 (𝑥)
= 4𝑥(1−𝑥2 ) = (1−𝑥2 ) , 0 < 𝑥 < 𝑦 < 1.

1 1 2𝑦 2(1−𝑥 3 )
Therefore, 𝐸[𝑌|𝑋 = 𝑥] = ∫𝑦=𝑥 𝑦. 𝑓𝑌|𝑋 (𝑦|𝑥) 𝑑𝑦 = ∫𝑦=𝑥 𝑦. (1−𝑥2 ) 𝑑𝑦 =
3(1−𝑥 2 )

2(1+𝑥+𝑥 2 )
= 3(1+𝑥)
, 𝑓𝑜𝑟 0 < 𝑥 < 1.

1 1
2𝑦
𝐸[𝑌 2 |𝑋 = 𝑥] = ∫ 𝑦 2 . 𝑓𝑌|𝑋 (𝑦|𝑥) 𝑑𝑦 = ∫ 𝑦2. 𝑑𝑦
𝑦=𝑥 𝑦=𝑥 (1 − 𝑥 2 )

(1+𝑥 2 )
= , 𝑓𝑜𝑟 0 < 𝑥 < 1.
2

𝑣𝑎𝑟[𝑌|𝑋 = 𝑥] = 𝐸[𝑌 2 |𝑋 = 𝑥] − (𝐸[𝑌|𝑋 = 𝑥])2


2
(1+𝑥 2 ) 2(1+𝑥+𝑥 2 ) (𝑥−1)2 (1+4𝑥+𝑥 2 )
= 2
−( 3(1+𝑥)
) = 18(1+𝑥)2
, 𝑓𝑜𝑟 0 < 𝑥 < 1.

1 1 2𝑦 1 2𝑦
Further, 𝐸[𝑋𝑌|𝑋 = 𝑥] = ∫𝑦=𝑥 𝑥𝑦. 𝑓𝑌|𝑋 (𝑦|𝑥) 𝑑𝑦 = ∫𝑦=𝑥 𝑥𝑦. (1−𝑥2 ) 𝑑𝑦 = 𝑥 ∫𝑦=𝑥 𝑦. (1−𝑥2 ) 𝑑𝑦

2𝑥(1+𝑥+𝑥 2 )
= 𝑥. 𝐸[𝑌|𝑋 = 𝑥] = 3(1+𝑥)
, 𝑓𝑜𝑟 0 < 𝑥 < 1.

Now the regression curve of 𝑌 𝑜𝑛 𝑋 = 𝑥 𝑖𝑠 𝑦 = 𝐸[𝑌|𝑋 = 𝑥]

2(1+𝑥+𝑥 2 )
i.e. 𝑦 = 3(1+𝑥)
, 𝑓𝑜𝑟 0 < 𝑥 < 1.

Exercises:

1. Find 𝑘 so that 𝑓(𝑥, 𝑦) = 𝑘(𝑥 + 𝑦), 0 < 𝑥 < 1, 0 < 𝑦 < 1 is a joint probability density function.

Ans. 1

2. The probability density function of a continuous bivariate distribution is given by

𝑥 + 𝑦, 𝑤ℎ𝑒𝑟𝑒 0 ≤ 𝑥 ≤ 1, 0 ≤ 𝑦 ≤ 1
𝑓(𝑥, 𝑦) = {
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒.

Find the marginal distributions and the correlation coefficient of X and Y.


1 1 1
Ans. 𝑓𝑋 (𝑥) = 𝑥 + , 0 ≤ 𝑥 ≤ 1; 𝑓𝑌 (𝑦) = 𝑦 + , 0 ≤ 𝑦 ≤ 1; 𝜌(𝑋, 𝑌) = − .
2 2 11

3. Let the two dimensional random variable (X, Y) have the joint density
1
𝑓(𝑥, 𝑦) = 8 (6 − 𝑥 − 𝑦)𝐼(0,2) (𝑥). 𝐼(2,4) (𝑦).

Find (i) 𝐸[𝑌|𝑋 = 𝑥]. (ii) 𝐸[𝑌 2 |𝑋 = 𝑥]. (iii) 𝑣𝑎𝑟[𝑌|𝑋 = 𝑥]. (iv) show that 𝐸[𝑌] = 𝐸[𝐸[𝑌|𝑋]].

(v) 𝐸[𝑋𝑌|𝑋 = 𝑥].

26−9𝑥 78−28𝑥 3𝑥 2 −18𝑥+26


Ans. (i) 3(3−𝑥) , 0 < 𝑥 < 2. (ii) 3(3−𝑥)
, 0 < 𝑥 < 2. (iii) 3(3−𝑥)2
, 0 < 𝑥 < 2.

𝑥(26−9𝑥)
(v) 3(3−𝑥)
, 0 < 𝑥 < 2.

You might also like