8th Material Subject: Joint Probability

Distribution (Discrete)
Undergraduate of Telecommunication Engineering
MUH1F3 - PROBABILITY AND STATISTICS

Telkom University
Center of eLearning & Open Education Telkom University
Jl. Telekomunikasi No.1, Bandung - Indonesia
[Link]

Lecturer: Nor Kumalasari Caecar Pratiwi, S.T., M.T. (caecarnkcp@[Link])

 TABLE OF CONTENTS:

1. Joint Probability Mass Functions

2. Marginal Probability Mass Functions
3. Conditional Probability Distribution
4. Covariance and Correlation

LEARNING OBJECTIVES:

After careful study of this chapter, student should be able to do the following:
1. Use joint probability mass functions to calculate probabilities
2. Calculate marginal and conditional probability distributions from joint probability distributions
3. Interpret and calculate covariance and correlations between random variables
2/15 May 10, 2020 LECTURER CODE: NKC/
 JOINT PROBABILITY MASS FUNCTION

For simplicity, we begin by considering random experiments in which only two random variables, called Bi-
[Link] Joint Probability Mass Function of the discrete random variables X and Y, denoted as fXY (xy),
satisfies:
fXY (xy) ≥ 0 (1)
XX
fXY (xy) = 1 (2)
X Y

fXY (xy) = P(X = x and Y = y) = P(X = x) ∩ P(Y = y) (3)

3/15 May 10, 2020 LECTURER CODE: NKC/

 MARGINAL PROBABILITY MASS FUNCTION

The Marginal Probability Mass Function of the discrete random variables X and Y, denoted as fX (x) or
fY (y), satisfies:
X
fX (x) = P(X = x) = fXY (x, y) (4)
y

X
fY (y) = P(Y = y) = fXY (x, y) (5)
x

4/15 May 10, 2020 LECTURER CODE: NKC/

 JOINT CUMULATIVE DISTRIBUTION FUNCTION

Remember that, for a random variable X, we define the CDF as FX (x) = P(X ≤ x). Now, if we have two ran-
dom variables X and Y and we would like to study them jointly, we can define the Joint Cumulative Function
as follows:
FXY (x, y) = P(X ≤ x and Y ≤ y) = P(X ≤ x) ∩ P(Y ≤ y) (6)

5/15 May 10, 2020 LECTURER CODE: NKC/

 INDEPENDENT BIVARIATE

The random variable X and Y become independent, if only:

fXY (x, y) = P(X = x) · P(Y = y) = fX (x) · fY (y) (7)

or:
FXY (x, y) = P(X ≤ x) · P(Y ≤ y) = FX (x) · FY (y) (8)

6/15 May 10, 2020 LECTURER CODE: NKC/

 COVARIANCE AND CORRELATION

When two random variables X and Y are not independent, it is frequently of interest to assess how strongly
they are related to one another. The Covariance between two random variables X and Y equal to:

Cov(XY) = E(XY) − E(X) · E(Y) (9)

Where, the joint expectation should be:

X
E(XY) = x · y · fXY (xy) (10)

The Correlation Coefficient of X and Y, equal to:

Cov(XY)
Cor(XY) = ρXY = (11)
σx · σy

7/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

Example: Will randomly pick two balls from a box that contains of three blue, two red and three green ball. If:
X = Random variables are declared elected as a blue ball
Y = Random variables are declared elected as a red ball

a. Determine range of random variable X

b. Determine range of random variable Y

c. Determine range of joint random variable X and Y

d. Determine the joint PMF of X and Y

e. Determine the marginal PMF of X

f. Determine the marginal PMF of X

g. Are the random variables X and Y independent?

h. If your answers are not independent, specify Cov(XY)] and ρXY

8/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

Answer:

a. Since X is a random variables declared elected as a blue ball, the range of X will:

RX = {0, 1, 2}

b. Since Y is a random variables declared elected as a red ball, the range of Y will:

RY = {0, 1, 2}

c. And range of joint random variable X and Y

RXY = {(0, 0), (0, 1), (0, 2), (1, 0), (1, 1), (2, 0)}

9/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

d. The joint PMF of X and Y

Suppose X = 0 and Y = 0, meaning that no blue or red balls are drawn. The two balls are taken from green
balls. So that:
3C0 · 2C0 · 3C2 3
fXY (0, 0) = =
8C2 28
While X = 0 and Y = 1, meaning that no blue drawn. The two balls are taken from 1 red and 1 green ball.
So that:
3C0 · 2C1 · 3C1 6
fXY (0, 1) = =
8C2 28
And X = 0 and Y = 2, meaning that the two balls are taken from red. So that:

3C0 · 2C2 · 3C0 1

fXY (0, 2) = =
8C2 28

In the same way, it can be calculated for fXY (1, 0), fXY (1, 1) and fXY (2, 0).

10/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

The joint PMF for X and Y shown below:

Y
0 1 2
0 3/28 6/28 1/28
X 1 9/28 6/28 0
2 3/28 0

e. The marginal PMF of X

3 6 1 10
fX (0) = + + =
28 28 28 28
9 6 15 3 3
fX (1) = + = and fX (2) = =
28 28 28 28 28

11/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

f. Determine the marginal PMF of Y

3 9 3 15 6 6 12 1
fY (0) = + + = , fY (1) = + = and fY (2) =
28 28 28 28 28 28 28 28

g. Random variables X and Y independent if fXY (xy) = fX (x) · fY (y)

fXY (0, 0) = fX (0) · fY (0)

3 10 15
6= ·
28 28 28
So, Random variables X and Y are not independent

12/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

i. The Cov(XY)] and ρXY are

10 15 3 21
X      
E(X) = x · fX (x) = 0· + 1· + 2· =
28 28 28 28

15 12 1 14
X      
E(Y) = y · fY (y) = 0· + 1· = + 2·
28 28 28 28
3 6 1 9
X        
E(XY) = x · y · fXY (xy) = 0 · 0 · + 0·1· + 0·2· + 1·0·
28 28 28 28
6 3 6
   
+ 1·1· + (1 · 2 · 0) + 2 · 0 · + (2 · 1 · 0) + (2 · 2 · 0) =
28 28 28
6 21 14 9
 
Cov(XY) = E(XY) − (E(X) · E(Y)) = − · =−
28 28 18 56

13/15 May 10, 2020 LECTURER CODE: NKC/

 EXAMPLE

10 15 3 27
X      
2 2 2 2 2
E(X )= x · fX (x) = 0 · + 1 · + 2 · =
28 28 28 28
 21 2 r
2 2 2 27 45 45
Var(X) = σ = E(X ) − (E(X)) =
x − = then σx =
28 28 112 112
15 12 1 16
X      
2 2 2 2 2
E(Y )= y · fY (y) = 0 · + 1 · + 2 · =
28 28 28 28
 21 2 r
2 2 2 14 9 9
Var(Y) = σ = E(Y ) − (E(Y)) =
y − = then σx =
28 28 112 112

− 569
ρXY = q q = −0.894
45 9
112
· 112

14/15 May 10, 2020 LECTURER CODE: NKC/

 Thank You

15/15 May 10, 2020 LECTURER CODE: NKC/