DISCRETE RANDOM VARIABLES

FAROUK MSELMI

INFERENTIAL STATISTICS

Farouk Mselmi DISCRETE RANDOM VARIABLES 1 / 52

PLAN

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 2 / 52

 Definitions and properties

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 3 / 52

 Definitions and properties

Definition
A discrete random variable is a function X (s) from a finite or countably infinite sample space S to
a subset A of N :
X (·) : S → A.

Example
Toss a coin 3 times in sequence. The sample space is
S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT }, and examples of random variables are
X (s) = the number of Heads in the sequence ; e.g., X (HTH) = 2,
Y (s) = The index of the first H ; e.g., Y (TTH) = 3,
Y (s) = 0 if the sequence has no H, i.e., Y (TTT ) = 0.

In this example, X (s) and Y (s) are actually integers.

Farouk Mselmi DISCRETE RANDOM VARIABLES 4 / 52

 Definitions and properties

Example
For the sample space S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT }, with X (s)
representing the number of Heads, the value X (s) = 2 corresponds to the event
{HHT , HTH, THH}, and the values 1 < X (s) ≤ 3 correspond to {HHH, HHT , HTH, THH}.

Notation
If it is clear what S is, then we often just write X instead of X (s).

Farouk Mselmi DISCRETE RANDOM VARIABLES 5 / 52

 Definitions and properties

Value-ranges of a random variable correspond to events in S, and events in S have a probability.

Thus, value-ranges of a random variable have a probability.

Example
For the sample space S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT }, with X (s)
representing the number of Heads, we have

6
P(0 < X ≤ 2) = .
8

Question
What are the values of P(X ≤ −1), P(X ≤ 0), P(X ≤ 1), P(X ≤ 2), P(X ≤ 3), P(X ≤ 4) ?

Farouk Mselmi DISCRETE RANDOM VARIABLES 6 / 52

 Definitions and properties

Example
For the sample space S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT }, with X (s)
representing the number of Heads, we have

1
P(X = 0) ≡ P({TTT }) = ,
8
3
P(X = 1) ≡ P({HTT , THT , TTH}) = ,
8
3
P(X = 2) ≡ P({HHT , HTH, THH}) = ,
8
1
P(X = 3) ≡ P({HHH}) = ,
8
where
P(X = 0) + P(X = 1) + P(X = 2) + P(X = 3) = 1. (Why ?)

Farouk Mselmi DISCRETE RANDOM VARIABLES 7 / 52

 Definitions and properties

The events E0 , E1 , E2 , E3 are disjoint since X (s) is a function ! ( X : S → A must be defined for all
s ∈ S and must be single-valued.)

Farouk Mselmi DISCRETE RANDOM VARIABLES 8 / 52

 Definitions and properties

Definition
P(X = x) is called the probability mass function.

Definition
FX (x) ≡ P(X ≤ x) is called the (cumulative) probability distribution function.

Properties
FX (x) is a non-decreasing function of x.
FX (−∞) = 0 and FX (∞) = 1.
P(a < X ≤ b) = FX (b) − FX (a).

Farouk Mselmi DISCRETE RANDOM VARIABLES 9 / 52

 Definitions and properties

Example
With X (s) = the number of Heads, and S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT },

1 3 3 1
P(X = 0) = , P(X = 1) = , P(X = 2) = , P(X = 3) = ,
8 8 8 8
we have the probability distribution function :

F (−1) ≡ P(X ≤ −1) = 0,

1
F (0) ≡ P(X ≤ 0) = ,
8
4
F (1) ≡ P(X ≤ 1) = ,
8
7
F (2) ≡ P(X ≤ 2) = ,
8
F (3) ≡ P(X ≤ 3) = 1,
F (4) ≡ P(X ≤ 4) = 1.

We see, for example, that

7 1 6
P(0 < X ≤ 2) = P(X = 1) + P(X = 2) = F (2) − F (0) = − = .
8 8 8

Farouk Mselmi DISCRETE RANDOM VARIABLES 10 / 52

 Definitions and properties

F IGURE: The graph of the cumulative distribution function FX .

Farouk Mselmi DISCRETE RANDOM VARIABLES 11 / 52

 Definitions and properties

Example
Toss a coin until ”Heads” occurs. Then the sample space is countably infinite, namely,

S = {H, TH, TTH, TTTH, . . .}.

The random variable X is the number of rolls until ”Heads” occurs :

X (H) = 1, X (TH) = 2, X (TTH) = 3, . . .

Then
1 1 1
P(X = 1) = , P(X = 2) = , P(X = 3) = , . . . (Why ?)
2 4 8
and
n n
X X 1 1
FX (n) = P(X ≤ n) = P(X = k ) = = 1− n,
2k 2
k =1 k =1

and, as should be the case,

∞ n  
X X 1
P(X = k ) = lim P(X = k ) = lim 1− = 1.
n→∞ n→∞ 2n
k =1 k =1

Farouk Mselmi DISCRETE RANDOM VARIABLES 12 / 52

 Definitions and properties Joint distributions

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 13 / 52

 Definitions and properties Joint distributions

Joint distributions

The probability mass function and the cumulative distribution function can also be functions of
more than one variable.
Example
Toss a coin 3 times in sequence. For the sample space
S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT }, we let

X (s) = # Heads, Y (s) = index of the first H (0 for TTT).

Then we have the joint probability mass function P(X = x, Y = y ). For example,

2 1
P(X = 2, Y = 1) = P(2 Heads, 1st toss is Heads) = = .
8 4
For S = {HHH, HHT , HTH, HTT , THH, THT , TTH, TTT }, X (s) = number of Heads, and Y (s) =
index of the first H, we can list the values of P(X = x, Y = y ) :

P(X = x, Y = y ) y =0 y =1 y =2 y =3 P(X = x)
x =0 1/8 0 0 0 1/8
x =1 0 1/8 1/8 1/8 3/8
x =2 0 2/8 1/8 0 3/8
x =3 0 1/8 0 0 1/8
P(Y = y ) 1/8 4/8 2/8 1/8 1

Farouk Mselmi DISCRETE RANDOM VARIABLES 14 / 52

 Definitions and properties Joint distributions

NOTE :
The marginal probability P(X = x) is the probability mass function of X .
The marginal probability P(Y = y ) is the probability mass function of Y .

Definitions
P(X = x, Y = y ) is called the joint probability mass function.
FX ,Y (x, y ) ≡ P(X ≤ x, Y ≤ y ) is called the joint (cumulative) probability distribution function.

Question : Why is

P(1 < X ≤ 3, 1 < Y ≤ 3) = F (3, 3) − F (1, 3) − F (3, 1) + F (1, 1) ?

Farouk Mselmi DISCRETE RANDOM VARIABLES 15 / 52

 Definitions and properties Independent random variables

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 16 / 52

 Definitions and properties Independent random variables

Independent random variables

Definition
Two discrete random variables X (s) and Y (s) are independent if

P(X = x, Y = y ) = P(X = x) · P(Y = y )

for all x and y .

The joint distribution function of independent random variables X and Y satisfies

FX ,Y (x, y ) = FX (x) · FY (y )

for all x, y .

Farouk Mselmi DISCRETE RANDOM VARIABLES 17 / 52

 Definitions and properties Conditional distributions

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 18 / 52

 Definitions and properties Conditional distributions

Conditional distributions

Definition
Let X and Y be discrete random variables with joint probability mass function P(X = x, Y = y ).
The conditional probability mass function P(X = x|Y = y ) is defined as

P(X = x, Y = y )
P(X = x|Y = y ) = ,
P(Y = y )

where P(X = x, Y = y ) is the joint probability mass function and P(Y = y ) is the marginal
probability mass function of Y .

Farouk Mselmi DISCRETE RANDOM VARIABLES 19 / 52

 Definitions and properties Expectation

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 20 / 52

 Definitions and properties Expectation

Expectation

The expected value of a discrete random variable X is denoted as E[X ], defined as

X
E[X ] ≡ xk · P(X = xk ).
k

Thus, E[X ] represents the weighted average value of X . ( E[X ] is also called the mean of X .)

Example
The expected value of rolling a die is

6
1 X 1 6·7 7
E[X ] = · k= · = .
6 6 2 2
k =1

Exercise : Prove the following :

E[aX ] = aE[X ]
E[aX + b] = aE[X ] + b

Farouk Mselmi DISCRETE RANDOM VARIABLES 21 / 52

 Definitions and properties Expectation

Example
Toss a coin until "Heads" occurs. The sample space is

S = {H, TH, TTH, TTTH, . . .}.

The random variable X is the number of tosses until "Heads" occurs :

X (H) = 1, X (TH) = 2, X (TTH) = 3, ...

The expected value of X is given by

n
1 1 2 1 3 1 1X k
E[X ] = · + · + · + . . . = lim .
2 2 4 2 8 2 n→∞ n 2k
k =1

n
1X k
The first few values of are as follows :
n 2k
k =1

1 Pn k
n n k =1 2k
1 0.50000000
2 1.00000000
3 1.37500000
10 1.98828125
40 1.99999999

Farouk Mselmi DISCRETE RANDOM VARIABLES 22 / 52

 Definitions and properties Expectation

The expected value of a function of a random variable is denoted as E[g(X )] and defined as
X
E[g(X )] ≡ g(xk ) · P(X = xk ).
k

Example
The pay-off of rolling a die is k 2 Dollar, where k is the side facing up. What should the entry fee be
for the betting to break even ? Here, g(X ) = X 2 , and

6
1X 2 1 6(6 + 1)(2 · 6 + 1) 91
E[g(X )] = k = · = ≈ $15.17.
6 6 6 6
k =1

Farouk Mselmi DISCRETE RANDOM VARIABLES 23 / 52

 Definitions and properties Expectation

Definition
The expected value of a function of two random variables is denoted as E[g(X , Y )] and defined as
XX
E[g(X , Y )] ≡ g(xk , y` ) · P(X = xk , Y = y` ).
k `

Properties
1 If X and Y are independent, then E[XY ] = E[X ] · E[Y ].
2 The expected value of the sum of random variables is given by E[X + Y ] = E[X ] + E[Y ].

Farouk Mselmi DISCRETE RANDOM VARIABLES 24 / 52

 Definitions and properties Variance and standard deviation

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 25 / 52

 Definitions and properties Variance and standard deviation

Variance and standard deviation

Let X have mean µ = E[X ]. Then, the variance of X is defined as

X
Var(X ) ≡ E[(X − µ)2 ] = (xk − µ)2 · P(X = xk ),
k

which is the average weighted square distance from the mean.

We have

Var(X ) = E[X 2 − 2µX + µ2 ] = E[X 2 ] − 2µE[X ] + µ2 = E[X 2 ] − 2µ2 + µ2 = E[X 2 ] − µ2 .

Farouk Mselmi DISCRETE RANDOM VARIABLES 26 / 52

 Definitions and properties Variance and standard deviation

The standard deviation of X is defined as

p q q
σ(X ) ≡ Var(X ) = E[(X − µ)2 ] = E[X 2 ] − µ2 ,

which is the square root of the variance and represents the average weighted distance from the
mean.

Example
The variance of rolling a die is

6
" #
X k2 · 1
Var(X ) = − µ2
6
k =1
 2
1 6(6 + 1)(2 · 6 + 1) 7
= · −
6 6 2
35
= .
12
The standard deviation is r
35
σ= ≈ 1.70.
12

Farouk Mselmi DISCRETE RANDOM VARIABLES 27 / 52

 Definitions and properties Covariance

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 28 / 52

 Definitions and properties Covariance

Covariance

Let X and Y be random variables with means E[X ] = µX , E[Y ] = µY . Then, the covariance of X
and Y is defined as
X
Cov(X , Y ) ≡ E[(X − µX )(Y − µY )] = (xk − µX )(y` − µY ) · P(X = xk , Y = y` ).
k ,`

We have

Cov(X , Y ) = E[(X − µX )(Y − µY )]

= E[XY − µX Y − µY X + µX µY ]
= E[XY ] − µX E[Y ] − µY E[X ] + µX µY
= E[XY ] − µX µY − µY µX + µX µY
= E[XY ] − E[X ]E[Y ].

NOTE : Cov(X , Y ) measures "concordance" or "coherence" of X and Y :

If X > µX when Y > µY and X < µX when Y < µY , then Cov(X , Y ) > 0.
If X > µX when Y < µY and X < µX when Y > µY , then Cov(X , Y ) < 0.

Farouk Mselmi DISCRETE RANDOM VARIABLES 29 / 52

 Definitions and properties Covariance

Exercise : Prove the following :

• Var(aX + b) = a2 Var(X ),
• Cov(X , Y ) = Cov(Y , X ),
• Cov(cX , Y ) = cCov(X , Y ),
• Cov(X , cY ) = cCov(X , Y ),
• Cov(X + Y , Z ) = Cov(X , Z ) + Cov(Y , Z ),
• Var(X + Y ) = Var(X ) + Var(Y ) + 2Cov(X , Y ).

Farouk Mselmi DISCRETE RANDOM VARIABLES 30 / 52

 Definitions and properties Covariance

Property
If X and Y are independent, then Cov(X , Y ) = 0.

PROOF : We have already shown (with µX ≡ E[X ] and µY ≡ E[Y ]) that

Cov(X , Y ) ≡ E[(X − µX )(Y − µY )]

= E[XY ] − E[X ]E[Y ],

and that if X and Y are independent, then E[XY ] = E[X ]E[Y ]. From which the result follows.

Farouk Mselmi DISCRETE RANDOM VARIABLES 31 / 52

 Definitions and properties Covariance

Property
If X and Y are independent, then Var(X + Y ) = Var(X ) + Var(Y ).

PROOF : We have

Var(X + Y ) = Var(X ) + Var(Y ) + 2Cov(X , Y ), (exercise)

and that if X and Y are independent, then Cov(X , Y ) = 0, from which the result follows.

Farouk Mselmi DISCRETE RANDOM VARIABLES 32 / 52

 Definitions and properties Moment-generating function

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 33 / 52

 Definitions and properties Moment-generating function

Moment-generating function

Definition
The moment-generating function of the random variable X is given by E(etX ) and is denoted by
MX (t). Hence, X
MX (t) = E(etX ) = etx P(X = x), if X is discrete.
x

Farouk Mselmi DISCRETE RANDOM VARIABLES 34 / 52

 Special discrete random variables

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 35 / 52

 Special discrete random variables The Bernoulli random variable

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 36 / 52

 Special discrete random variables The Bernoulli random variable

The Bernoulli random variable

A Bernoulli trial has only two outcomes, with probability

P(X = 1) = p, P(X = 0) = 1 − p,

e.g., tossing a coin, winning or losing a game, . . .. We have

E[X ] = 1 · p + 0 · (1 − p) = p,

E[X 2 ] = 12 · p + 02 · (1 − p) = p,
Var(X ) = E[X 2 ] − (E[X ])2 = p − p2 = p(1 − p).
NOTE : If p is small, then Var(X ) ≈ p.

Farouk Mselmi DISCRETE RANDOM VARIABLES 37 / 52

 Special discrete random variables The Bernoulli random variable

Example
1
When p = 2
(e.g., for tossing a coin), we have

1
E[X ] = p = ,
2
1
Var(X ) = p(1 − p) = .
4
When rolling a die with outcome k (1 ≤ k ≤ 6), let
(
1, if the roll resulted in a six,
X (k ) =
0, if the roll did not result in a six.

Then
1
E[X ] = p = ,
6
5
Var(X ) = p(1 − p) = .
36
When p = 0.01, then
E[X ] = 0.01,
Var(X ) = 0.0099 ≈ 0.01.

Farouk Mselmi DISCRETE RANDOM VARIABLES 38 / 52

 Special discrete random variables The Binomial random variable

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 39 / 52

 Special discrete random variables The Binomial random variable

The Binomial random variable

Perform a Bernoulli trial n times in sequence. Assume the individual trials are independent. An
outcome could be 100011001010 (with n = 12), with probability

P(100011001010) = p5 · (1 − p)7 .

Let X be the number of "successes" (i.e., 1’s). For example, X (100011001010) = 5. We have
12
P(X = 5) = · p5 · (1 − p)7 .
5
In general, for k successes in a sequence of n trials, we have
n 
P(X = k ) = · pk · (1 − p)n−k ,
k
where 0 ≤ k ≤ n.

Farouk Mselmi DISCRETE RANDOM VARIABLES 40 / 52

 Special discrete random variables The Binomial random variable

Example
In 12 rolls of a die, write the outcome as, for example, 100011001010, where 1 denotes the roll
resulted in a six, and 0 denotes the roll did not result in a six. As before, let X be the number of 1’s
in the outcome. Then X represents the number of sixes in the 12 rolls. Then,

P(X = 5) ≈ 2.8%, P(X ≤ 5) ≈ 99.2%.

Farouk Mselmi DISCRETE RANDOM VARIABLES 41 / 52

 Special discrete random variables The Binomial random variable

Mean and variance of the Binomial random variable

By definition, the mean of a Binomial random variable X is

n
X n
X n 
E[X ] = k · P(X = k ) = k· · pk · (1 − p)n−k ,
k
k =0 k =0

which can be shown to equal np. An easy way to see this is as follows : If in a sequence of n
independent Bernoulli trials we let

Xk = the outcome of the k th Bernoulli trial,

(Xk = 0 or 1),
then
X ≡ X1 + X2 + . . . + Xn ,
is the Binomial random variable that counts the "successes".

Farouk Mselmi DISCRETE RANDOM VARIABLES 42 / 52

 Special discrete random variables The Binomial random variable

X ≡ X1 + X2 + . . . + Xn
We know that
E[Xk ] = p,
so
E[X ] = E[X1 ] + E[X2 ] + . . . + E[Xn ] = np.
We already know that

Var(Xk ) = E[Xk2 ] − (E[Xk ])2 = p − p2 = p(1 − p),

so, since the Xk are independent, we have

Var(X ) = Var(X1 ) + Var(X2 ) + . . . + Var(Xn ) = np(1 − p).

NOTE : If p is small, then Var(X ) ≈ np.

Farouk Mselmi DISCRETE RANDOM VARIABLES 43 / 52

 Special discrete random variables The Binomial random variable

Example
1
For 12 tosses of a coin, with Heads as success, we have n = 12, p = 2
, so

E[X ] = np = 6,

Var(X ) = np(1 − p) = 3.
1
For 12 rolls of a die, with six as success, we have n = 12, p = 6
, so

E[X ] = np = 2,

5
Var(X ) = np(1 − p) = .
3
If n = 500 and p = 0.01, then
E[X ] = np = 5,
Var(X ) = np(1 − p) = 4.95 ≈ 5.

Farouk Mselmi DISCRETE RANDOM VARIABLES 44 / 52

 Special discrete random variables The Poisson random variable

1 Definitions and properties

Joint distributions
Independent random variables
Conditional distributions
Expectation
Variance and standard deviation
Covariance
Moment-generating function

2 Special discrete random variables

The Bernoulli random variable
The Binomial random variable
The Poisson random variable

Farouk Mselmi DISCRETE RANDOM VARIABLES 45 / 52

 Special discrete random variables The Poisson random variable

The Poisson random variable

The Poisson variable approximates the Binomial random variable :

n  λk
P(X = k ) = · pk · (1 − p)n−k ≈ e−λ · ,
k k!
when we take λ = np (the average number of successes). This approximation is accurate if n is
large and p is small.
Recall that for the Binomial random variable :

E[X ] = np,

and
Var(X ) = np(1 − p) ≈ np when p is small.
Indeed, for the Poisson random variable, we will show that :

E[X ] = λ,

and
Var(X ) = λ.

Farouk Mselmi DISCRETE RANDOM VARIABLES 46 / 52

 Special discrete random variables The Poisson random variable

A stable and efficient way to compute the Poisson probability

λk
P(X = k ) = e−λ · ,
k!
for k = 0, 1, 2, . . ., is to use the recurrence relation

P(X = 0) = e−λ ,

λ
P(X = k + 1) = · P(X = k ),
k +1
for k = 0, 1, 2, . . ..
NOTE : Unlike the Binomial random variable, the Poisson random variable can have an arbitrarily
large integer value k .

Farouk Mselmi DISCRETE RANDOM VARIABLES 47 / 52

 Special discrete random variables The Poisson random variable

The Poisson random variable

λk
P(X = k ) = e−λ · ,
k!
for k = 0, 1, 2, . . ., has (as shown later) :

E[X ] = λ

and
Var(X ) = λ.
The Poisson distribution function is
k k
X λ` X λ`
FX (k ) = P(X ≤ k ) = e−λ · = e−λ ,
`! `!
`=0 `=0

with, as should be the case,

∞
X λ`
lim FX (k ) = e−λ = e−λ eλ = 1.
k →∞ `!
`=0

(using the Taylor series from Calculus for eλ ).

Farouk Mselmi DISCRETE RANDOM VARIABLES 48 / 52

 Special discrete random variables The Poisson random variable

The Poisson random variable

λk
P(X = k ) = e−λ · ,
k!
for k = 0, 1, 2, . . ., models the probability of k "successes" in a given "time" interval, when the
average number of successes is λ.

Example
Suppose customers arrive at the rate of six per hour. The probability that k customers arrive in a
one-hour period is
60
P(k = 0) = e−6 · ≈ 0.0024,
0!
61
P(k = 1) = e−6 · ≈ 0.0148,
1!
62
P(k = 2) = e−6 · ≈ 0.0446.
2!
The probability that more than 2 customers arrive is

1 − (0.0024 + 0.0148 + 0.0446) ≈ 0.938.

Farouk Mselmi DISCRETE RANDOM VARIABLES 49 / 52

 Special discrete random variables The Poisson random variable

The method of moments

By Taylor expansion of etX about t = 0, we have

" #
tX t 2X 2 t 3X 3 t2 t3
MX (t) ≡ E[e ] = E 1 + tX + + + . . . = 1 + tE[X ] + E[X 2 ] + E[X 3 ] + . . . .
2! 3! 2! 3!

It follows that
MX0 (0) = E[X ], MX00 (0) = E[X 2 ].
This sometimes facilitates computing the mean µ = E[X ], and the variance Var(X ) = E[X 2 ] − µ2 .

Farouk Mselmi DISCRETE RANDOM VARIABLES 50 / 52

 Special discrete random variables The Poisson random variable

Application : The Poisson mean and variance

∞ ∞ ∞
X X λk X (λet )k t t
MX (t) ≡ E[etX ] = etk P(X = k ) = etk e−λ = e−λ = e−λ eλe = eλ(e −1) .
k! k!
k =0 k =0 k =0

Here,
t
MX0 (t) = λet eλ(e −1) ,

t
MX00 (t) = λ λe2t + et eλ(e −1) .

So that,
E[X ] = MX0 (0) = λ,
E[X 2 ] = MX00 (0) = λ(λ + 1) = λ2 + λ,
Var(X ) = E[X 2 ] − E[X ]2 = λ.

Farouk Mselmi DISCRETE RANDOM VARIABLES 51 / 52

 Special discrete random variables The Poisson random variable

THANK YOU

Farouk Mselmi DISCRETE RANDOM VARIABLES 52 / 52