School of Commercial High Studies

Ecole des Hautes Etudes Commerciales (EHEC)

Probabilty courses
2nd year of preparatory classes

Chapter I : Random variables

and probability distributions
Chapitre I : Variables aléatoires
et distributions de probabilité

Prepared by: ALLOUAT Asma

Academic year: 2024/2025
 Section 2 :
Discrete random variables
Section 2 : Les variables aléatoires discrètes
 3
Discrete random variables

Plan:
1. Probability Mass Function and Cumulative Distribution Function
2. Moments of a random variable
3. Generating functions
4. Transformations of random variables
5. Markov and Chebyshev inequalities
1. Probability Mass Function (PMF) 4
and Cumulative Distribution Function (CDF)

Introductory example

Consider the experiment consisting of rolling an unfair (non

balanced) die.
We define the random variable 𝑋 that assigns the obtained
number 𝑥i.
the number on which the die comes to rest
 5
The range :
𝑋(Ω) = {1, 2, 3, 4, 5, 6}
Let fX be a function defined by: fX (𝑥i) = 𝑃(𝑋=𝑥i) = 𝑥i / 21

The probability distribution of 𝑋 is given by:

𝑥i 1 2 3 4 5 6 ∑
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21 1

The function f X (𝑥i) is called Probability mass function (fonction

de masse)
 𝑥i 1 2 3 4 5 6
6
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21

Let define the function: FX (𝑥i) = 𝑃(𝑋 ≤ 𝑥i)

𝑥i 1 2 3 4 5 6
𝑃(𝑋≤ 𝑥𝑥ii))
𝑃(𝑋≤ 1/21 3/21 6/21 10/21 15/21 21/21

The fonction FX(𝑥i) is called Cumulative distribtion function or

CDF( fonction de répartition)
𝟎 𝒊𝒇 𝒙𝒊 < 𝟏
𝟏 𝟐𝟏 𝒊𝒇 𝟏 ≤ 𝒙𝒊 < 𝟐
𝟑 𝟐𝟏 𝒊𝒇 𝟐 ≤ 𝒙𝒊 < 𝟑
𝑭𝑿 𝒙𝒊 = 𝟔 𝟐𝟏 𝒊𝒇 𝟑 ≤ 𝒙𝒊 < 𝟒
𝟏𝟎 𝟐𝟏 𝒊𝒇 𝟒 ≤ 𝒙𝒊 < 𝟓
𝟏𝟓 𝟐𝟏 𝒊𝒇 𝟓 ≤ 𝒙𝒊 < 𝟔
𝟏 𝒊𝒇 𝒙𝒊 ≥ 𝟔
 7
1.1. Probability function or Probability Mass Function - PMF
(Fonction de masse )

Definition:
For a discrete random variable X, the probability mass
function is the function fX(𝑥i) that assigns to each realization
𝑥i ∊ 𝑋(Ω), a probability 𝑃(𝑋=𝑥i) :

fX(𝑥i) = 𝑃(𝑋=𝑥i) ∀𝑥i ∊ 𝑋(Ω)

1.2. Distribution function or Cumulative Distribution 8
Function -CDF (Fonction de répartition)

Definition1
The cumulative distribution function or briefly the distribution
function 𝐹X(𝑥), for a random variable X is given by:

∀ 𝑥 ∈ ℝ, 𝐹𝑋 (𝑥) = 𝑃(𝑋 ≤ 𝑥)
 9
Definition 2
For a discrete random variableX ,the distribution function de is
the mapping 𝐹𝑋 such as :

𝐹𝑋 (𝑥) : ℝ → [0,1]
𝑥 ↦ 𝑭𝑿 (𝒙) = 𝐏(𝐗 ≤ 𝒙) = 𝒙𝒊 ∊𝑿 𝜴 , 𝒙𝒊 ≤𝒙 𝑷(𝑿 = 𝒙𝒊 )

Other general notation:

𝒙

𝑭𝑿 𝒙 = 𝑷 𝑿=𝒕 with 𝒕 ∊ 𝑿 𝜴 ,
𝒕=−∞
 10
Properties :
From this definition, several properties of the CDF can be
inferred :

 𝐹𝑋 is non-decreasing
 𝐹𝑋 is right-continuous

 lim F𝑋 𝑥 = 0
𝑥→ −∞

 lim F𝑋 (𝑥) = 1
𝑥→ +∞
Probabilities calculation 11
Probabilities of events for a variable X can be given in terms
of its distribution function 𝐹𝑋 (𝑥).
∀ (𝑥i , 𝑥j) ∊ℝ2 with 𝑥j > 𝑥i we have:

 𝑃(𝑥i < 𝑋 ≤ 𝑥j ) = 𝐹𝑋(𝑥j ) −𝐹𝑋 (𝑥i)

 𝑃(𝑥i ≤ 𝑋 ≤ 𝑥j ) = 𝐹𝑋(𝑥j ) −𝐹𝑋 (𝑥i-1)
 𝑃(𝑥i ≤ 𝑋 < 𝑥j ) = 𝐹𝑋(𝑥j -1 ) −𝐹𝑋 (𝑥i -1)
 𝑃(𝑥i < 𝑋 < 𝑥j ) = 𝐹𝑋(𝑥j -1 ) −𝐹𝑋 (𝑥i)
 𝑃(𝑋 > 𝑥i ) = 1−𝑃(𝑋 ≤ 𝑥i ) =1 −𝐹𝑋(𝑥i )
 𝑃(𝑋 ≥ 𝑥i ) = 1−𝑃(𝑋 < 𝑥i ) =1 − 𝐹𝑋 (𝑥i -1)
 𝑃(𝑋 = 𝑥i ) = 𝐹𝑋(𝑥i ) − 𝐹𝑋(𝑥i-1 )
 In the previous example:
𝑥i 1 2 3 4 5 6 ∑ 12
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21 1
𝐹𝑋 (𝑥i) =𝑃(𝑋≤ 𝑥i) 1/21 3/21 6/21 10/21 15/21 1 -
𝟏𝟓 𝟑 𝟏𝟐
 𝑃(𝑥i < 𝑋 ≤ 𝑥j ) = 𝐹𝑋(𝑥j ) −𝐹𝑋 (𝑥i)  𝑃( 2 < 𝑋 ≤ 5 ) = 𝐹𝑋(5 ) −𝐹𝑋 (2) = − =
𝟐𝟏 𝟐𝟏 𝟐𝟏

𝟏𝟓 𝟏 𝟏𝟒
 𝑃(𝑥i ≤ 𝑋 ≤ 𝑥j ) = 𝐹𝑋(𝑥j ) −𝐹𝑋 (𝑥i-1)  𝑃( 2 ≤ 𝑋 ≤ 5 ) = 𝐹𝑋(5 ) −𝐹𝑋 (1) = − =
𝟐𝟏 𝟐𝟏 𝟐𝟏

𝟏𝟎 𝟏 𝟗
 𝑃(𝑥i ≤ 𝑋 < 𝑥j ) = 𝐹𝑋(𝑥j -1 ) −𝐹𝑋 (𝑥i -1)  𝑃( 2 ≤ 𝑋 < 5) = 𝐹𝑋(4) −𝐹𝑋 (1) = − =
𝟐𝟏 𝟐𝟏 𝟐𝟏

𝟏𝟎 𝟑 𝟕
 𝑃(𝑥i < 𝑋 < 𝑥j ) = 𝐹𝑋(𝑥j -1 ) −𝐹𝑋 (𝑥i)  𝑃( 2 < 𝑋 < 5) = 𝐹𝑋(4) −𝐹𝑋 (2) = − =
𝟐𝟏 𝟐𝟏 𝟐𝟏

𝟏𝟎 𝟏𝟏
 𝑃(𝑋 > 𝑥i ) = 1−𝑃(𝑋 ≤ 𝑥i ) =1 −𝐹𝑋(𝑥i )  𝑃(𝑋 > 4 ) = 1−𝑃(𝑋 ≤ 4 ) =1 −𝐹𝑋(4 ) = 𝟏 − =
𝟐𝟏 𝟐𝟏

𝟔 𝟏𝟓
 𝑃(𝑋 ≥ 𝑥i ) = 1−𝑃(𝑋 < 𝑥i ) =1 − 𝐹𝑋 (𝑥i -1)  𝑃(𝑋 ≥ 4 ) = 1−𝑃(𝑋 < 4 ) =1 −𝐹𝑋(3 ) = 𝟏 − =
𝟐𝟏 𝟐𝟏

𝟏𝟎 𝟔 𝟒
 𝑃(𝑋 = 𝑥i ) = 𝐹𝑋(𝑥i ) − 𝐹𝑋(𝑥 i-1 )  𝑃(𝑋 = 4 ) = 𝐹𝑋(4 ) − 𝐹𝑋(3) = − =
𝟐𝟏 𝟐𝟏 𝟐𝟏
 13
2. Moments of a random variable
(Moments d’une variable aléatoire)

2.1. Expectation and variance

 (Mathematical) expectation (l’espérance mathématique)
 Variance (la variance)

2.2. Other moments

 Raw moments (les moments simples ou ordinaires (non
centrés)
 Central Moments (les moments centrés)
 Factorial moments (les moments factoriels)
2.1. Expectation and variance 14

 (Mathematical) expectation, expected value or mean

(L’espérance mathématique)

Définition
Let 𝑋 be a discrete random variable defined on 𝑋(Ω) by
the mass function 𝑃(𝑋=𝑥) ,
The expected value of 𝑋, denoted E(𝑋) or µ is defined by:
𝒏

𝝁=𝐄 𝑿 = 𝒙𝒊 𝐏 𝑿 = 𝒙𝒊 𝒙𝒊 ∊ 𝑿(𝛀)
𝒊=𝟏
Example 15

In the previous example of rolling a die:

𝑥i 1 2 3 4 5 6 ∑
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21 1
𝒏

𝐄 𝑿 = 𝒙𝒊 𝐏 𝑿 = 𝒙 𝒊
𝒊=𝟏

𝟏 𝟐 𝟑 𝟒 𝟓 𝟔 𝟏𝟑
𝐄 𝐗 =𝟏× + 𝟐× + 𝟑× + 𝟒× + 𝟓× + 𝟔× = = 𝟒, 𝟑𝟑
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟑
 16
Expected value of a function
Let 𝑋 be a discrete random variable defined on 𝑋(Ω)
by the mass function 𝑃(𝑋=𝑥) ,

The expected value of the function ℎ(𝑋), noted 𝐸[ℎ(𝑋)]

where ℎ(.) is a measurable function, is defined by:

𝐄 (𝒉(𝐗)) = 𝒉(𝒙𝒊 ) 𝐏 𝐗 = 𝒙𝒊 𝒙𝒊 ∊ 𝐗(𝛀)

𝒊=𝟏
Example 17

In the previous example of rolling a die:

𝑥i 1 2 3 4 5 6 ∑
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21 1

Let ℎ(𝑋) be a function

𝒉(𝐗) = X 2
𝒏

𝐄 (𝒉(𝐗)) = 𝐄 𝑿𝟐 = 𝒙 𝒊 𝟐 𝐏 𝑿 = 𝒙𝒊
𝒊=𝟏
𝟏 𝟐 𝟑 𝟒 𝟓 𝟔
𝐄 𝑿𝟐 =𝟏× +𝟒× +𝟗× + 𝟏𝟔 × + 𝟐𝟓 × + 𝟑𝟔 × = 𝟐𝟏
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
 18
Properties

 E (𝑋+ a ) = E (𝑋)+ a ∀ a ∊ℝ

 E (a 𝑋 ) = a E (𝑋)

 E (𝑋+Y ) = E (𝑋) + E (Y)

E (a 𝑋+ b Y ) = a E (𝑋)+ b E (Y) ∀ (a, b) ∊ℝ2

 19
 The variance (la variance)

The variance, denoted 𝑉(𝑋), of the discrete random variable 𝑋

defined on 𝑋(Ω) is given by:

𝐕 𝐗 = 𝑬 (𝒙𝒊 − 𝐄 𝐗 )𝟐
Then:
𝒏

𝐕(𝐗) = (𝒙𝒊 − 𝐄 𝐗 )𝟐 𝐏(𝐗 = 𝒙𝒊 )

𝒊=𝟏
Example
20
In the previous example:

𝑥i 1 2 3 4 5 6 ∑
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21 1
𝒏

𝐕(𝐗) = 𝑬 (𝒙𝒊 − 𝐄 𝐗 )𝟐 = (𝒙𝒊 − 𝐄 𝐗 )𝟐 𝐏(𝐗 = 𝒙𝒊 )

𝒊=𝟏

𝟏
𝟐 𝟐
𝟐 𝟐
𝟑
𝐕 𝐗 = 𝟏 − 𝟒, 𝟑𝟑 × + 𝟐 − 𝟒, 𝟑𝟑 × + 𝟑 − 𝟒, 𝟑𝟑 ×
𝟐𝟏 𝟐𝟏 𝟐𝟏
𝟐
𝟒 𝟐
𝟓 𝟐
𝟔
+ 𝟒 − 𝟒, 𝟑𝟑 × + 𝟓 − 𝟒, 𝟑𝟑 × + 𝟔 − 𝟒, 𝟑𝟑 × = 𝟐, 𝟐𝟐
𝟐𝟏 𝟐𝟏 𝟐𝟏
* Standard deviation (L’écart-type) 21

For a discrete random variable 𝑋 defined on 𝑋(Ω),

The standard deviation 𝜎 (𝑋) is defined by :

𝜎(𝑋) = 𝐕(𝐗)

Example:
In the previous example: 𝑽(𝑿) = 𝟐, 𝟐𝟐

𝝈(𝑿) = 𝟐, 𝟐𝟐 = 𝟏, 𝟒𝟗
Properties 22

 V (𝑋+ a ) = V (𝑋) ∀ a ∊ℝ

 V (a 𝑋 ) = a2 V (𝑋)

 V (a 𝑋 + b ) = a 2 V (𝑋) ∀ (a, b) ∊ℝ2

König-Huygens formula 23

The general expression 𝐕 𝐗 = 𝐄 (𝐗 − 𝐄(𝐗))𝟐 can be developed

and written as:

𝐕 𝐗 = 𝐄 𝐗 𝟐 − (𝐄(𝐗))𝟐

Example:
In the previous example, we found 𝑽(𝑿) = 𝟐, 𝟐𝟐
We can find the same result using König-Huygens form:
𝟐
𝟐
𝟏𝟑 𝟐𝟎
𝑽 𝑿 =𝑬 𝑿 − (𝑬 𝑿 )𝟐 = 𝟐𝟏 − = = 𝟐, 𝟐𝟐
𝟑 𝟗
 24
2.2. Other moments
(Autres moments)

Let 𝑋 be a discrete random variable defined on 𝑋(Ω) by the

mass function 𝑃(𝑋=𝑥) ,

The random variable X, can be characterised by its:

 Raw moments (moments simples ou ordinaires)

 Central Moments (moments centrés)
 Factorial moments (moments factoriels)
Raw moments (Moments ordinaires ) 25
The k th order raw moment of the random variable 𝑋 ( k ∊ N) is defined by:
𝒏

𝒎𝒌 = 𝐄 𝐗 𝒌 = 𝒙𝒌𝒊 𝐏(𝐗 = 𝒙𝒊 )
𝒊=𝟏
Example
The 1st order raw moment in the rolling die example is given by :

𝟏 𝟐 𝟑 𝟒 𝟓 𝟔
𝒎𝟏 = 𝐄 𝐗𝟏 = 𝟏𝟏 × 𝟏
+ 𝟐 × 𝟏
+ 𝟑 × 𝟏
+ 𝟒 × 𝟏
+ 𝟓 × 𝟏
+ 𝟔 × = 𝟒, 𝟑𝟑
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏

The 2nd order raw moment is given by :

𝟐
𝟏 𝟐 𝟐
𝟐 𝟐
𝟑 𝟐
𝟒 𝟐
𝟓 𝟐
𝟔
𝒎𝟐 = 𝐄 𝐗 =𝟏 × + 𝟐 × + 𝟑 × + 𝟒 × + 𝟓 × + 𝟔 × = 𝟐𝟏
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
Central Moments (Moments centrés ) 26
The k th order central moment of the random variable 𝑋 ( k ∊ N) is defined by:
𝒏
𝐤
𝝁𝒌 = 𝐄 (𝐗 − 𝐄(𝐗)) = ( 𝒙𝒊 − 𝐄 𝐗 )𝐤 𝐏(𝐗 = 𝒙𝒊 )
𝒊=𝟏
Example
The 1st order central moment in the rolling die example is given by :
𝟏 𝟏
𝟏 𝟏
𝟐 𝟏
𝟑
𝝁𝟏 = 𝐄 (𝐗 − 𝐄(𝐗)) = 𝟏 − 𝟒, 𝟑𝟑 × + 𝟐 − 𝟒, 𝟑𝟑 × + 𝟑 − 𝟒, 𝟑𝟑 ×
𝟐𝟏 𝟐𝟏 𝟐𝟏
𝟏
𝟒 𝟏
𝟓 𝟏
𝟔
+ 𝟒 − 𝟒, 𝟑𝟑 × + 𝟓 − 𝟒, 𝟑𝟑 × + 𝟔 − 𝟒, 𝟑𝟑 × = 𝟎
𝟐𝟏 𝟐𝟏 𝟐𝟏

The 2nd order central moment is given by :

𝟐 𝟏 𝟐 𝟑
𝝁𝟐 = 𝐄 (𝐗 − 𝐄(𝐗)) = 𝟏 − 𝟒, 𝟑𝟑 𝟐 × + 𝟐 − 𝟒, 𝟑𝟑 𝟐 × + 𝟑 − 𝟒, 𝟑𝟑 𝟐 ×
𝟐𝟏 𝟐𝟏 𝟐𝟏
𝟐
𝟒 𝟐
𝟓 𝟐
𝟔
+ 𝟒 − 𝟒, 𝟑𝟑 × + 𝟓 − 𝟒, 𝟑𝟑 × + 𝟔 − 𝟒, 𝟑𝟑 × = 𝟐, 𝟐𝟐
𝟐𝟏 𝟐𝟏 𝟐𝟏
 27
Factorial moments (Moments factoriels)

The k th order factorial moment of the random variable 𝑋 ( k ∊ N) is defined by:

𝒏

𝐄 𝑿 𝐗 − 𝟏 … (𝐗 − 𝐤 + 𝟏) = 𝑿 𝐗 − 𝟏 … (𝐗 − 𝐤 + 𝟏) 𝐏(𝐗 = 𝒙𝒊 )
𝒊=𝟏
Example

The 2nd order factorialmoment in the rolling die example is given by :

𝟏 𝟐 𝟑
𝐄 𝑿 𝐗−𝟏 = 𝟏 𝟏−𝟏 + 𝟐 𝟐−𝟏 + 𝟑 𝟑−𝟏
𝟐𝟏 𝟐𝟏 𝟐𝟏
𝟒 𝟓 𝟔
+ 𝟒 𝟒−𝟏 + 𝟓 𝟓−𝟏 + 𝟔 𝟔−𝟏 = 𝟏𝟔, 𝟔𝟕
𝟐𝟏 𝟐𝟏 𝟐𝟏
 28

Remarks :
The expectation and the variance are special moments :

 The 1st order raw moment and the 1st order

factorial moment represent the expectation.

 The 2nd order central moment is the variance.

3. Generating functions 29
(Fonctions génératrices )

The generating functions are tools used generally for

finding moments, probabilities or the distribution of
sums of random variables.

- Moments generating function

(La fonction génératrice des moments)

- Probability generating function

(La fonction génératrice des probabilités)
3.1. Moments generating function 30
(La fonction génératrice des moments)

Definition:
Let 𝑋 be a discrete random variable defined on 𝑋(Ω) by the
mass function 𝑃(𝑋=𝑥)
The moment generating function (MGF) of X, denoted 𝑴𝑿 𝒕 ,
is defined for all values of t by:

𝑴𝑿 𝒕 = 𝑬 𝒆𝒕𝑿 = 𝒙𝒊 ∈𝑿(𝛀) 𝒆𝒕𝒙𝒊 𝑷(𝑿 = 𝒙 )

𝒊
Properties:
* Let 𝑋 be a discrete random variable and let 𝑴𝑿 𝒕 be its 31
moment generating function, we have:
𝝏𝒌𝑴 𝟎
𝒌 (𝒌) 𝑿
𝐄[𝑿 ] = 𝐌𝐗 𝟎 =
(𝝏𝒕)𝒌
Thus:
𝝏 𝑴𝑿 𝟎
𝐄𝑿 = 𝐌𝐗′ 𝟎 =
𝝏𝒕

𝝏𝟐 𝑴𝑿 𝟎
𝐄[𝑿𝟐 ] = 𝐌𝐗′′ 𝟎 =
(𝝏𝒕)𝟐
We notice that for k = 0 :
𝑴𝑿 𝟎 = 𝑬 𝒆𝟎.𝑿 = 𝐄[𝟏] = 𝟏
 32
* The moment generating function of the sum of
independent random variables is the product of the
individual moment generating functions

* If 2 random variables X and Y have the same moment

generating function (i.e. M𝑋 (t) = M𝑌(t)), then: 𝑋 et 𝑌 have
the same probability distribution.

* If 𝑌=𝑎 𝑋+𝑏, then: 𝑴𝒀 𝒕 = 𝒆𝒃𝒕 𝑴𝑿 𝒂𝒕

Example : 33
In the previous example of the die.

The moment generating function M𝑋(𝑡) is given by:

𝟏 𝒕 𝟐 𝟐𝒕 𝟑 𝟑𝒕 𝟒 𝟒𝒕 𝟓 𝟓𝒕 𝟔 𝟔𝒕
𝑴𝑿 𝒕 = 𝒆 + 𝒆 + 𝒆 + 𝒆 + 𝒆 + 𝒆
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏

The raw moments are obtained as follows:

1 𝒕 4 2𝒕 9 3𝒕 16 4𝒕 25 5𝒕 36 6𝒕 𝐄[𝐗] = 𝐌𝐗′ 0 = 4,33
𝐌𝐗′ 𝒕 = 𝒆 + 𝒆 + 𝒆 + 𝒆 + 𝒆 + 𝒆
21 21 21 21 21 21

1 𝒕 8 2𝒕 27 3𝒕 64 4𝒕 125 5𝒕 216 6𝒕 𝐄[𝐗 2 ] = 𝐌𝐗′′ 0 = 21

𝐌𝐗′′ 𝒕 = 𝒆 + 𝒆 + 𝒆 + 𝒆 + 𝒆 + 𝒆
21 21 21 21 21 21
3.2. Probability Generating Function 34
(La fonction génératrice des probabilités)

The PGF is used to obtain the probabilities for each value 𝑥i ∈𝑋(Ω)
of a discrete random variable.

Définition
The Probability Generating Function denoted 𝐺𝑋(𝑡), of a discrete
random 𝑋 non negative (𝑥 ∈{0,1,2,…}) is defined by:

𝑮 𝑿 𝒕 = 𝑬 𝒕𝑿 = 𝒕𝒙𝒊 𝑷 𝑿 = 𝒙𝒊
𝒙𝒊 ∈𝑿(𝛀)
Example :
35
In the previous example of the die,

𝑥i 1 2 3 4 5 6 ∑
𝑃(𝑋=𝑥i) 1/21 2/21 3/21 4/21 5/21 6/21 1

The Probability Generating function 𝐺𝑋(𝑡) is given by:

𝑮𝑿 𝒕 = 𝑬 𝒕 𝑿 = 𝒕 𝒙𝒊 𝑷 𝑿 = 𝒙 𝒊
𝒙𝒊 ∈𝑿(𝛀)

𝟏 𝟐 𝟐 𝟑 𝟑 𝟒 𝟒 𝟓 𝟓 𝟔 𝟔
𝑮𝑿 𝒕 = 𝒕+ 𝒕 + 𝒕 + 𝒕 + 𝒕 + 𝒕
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
Property1: 36

Let 𝐺𝑋(𝑡) be the Probability Generating Function of a

discrete random variable X defined on 𝑋(Ω),
(𝒌)
𝐆𝐗 𝟎
We have : 𝑷 𝑿=𝒌 =
𝒌!

Particularly: 𝑷 𝑿 = 𝟎 = 𝑮𝑿 𝟎

𝑷 𝑿 = 𝟏 = 𝑮′𝑿 𝟎

𝟏 ′′
𝑷 𝑿=𝟐 = 𝑮
𝟐 𝑿
𝟎

…
Example : 37
Let 𝐺𝑋(𝑡) be the Probability Generating Function defined by:
𝟏 𝟑 𝟏
𝑮𝑿 𝒕 = 𝒕 + 𝒕+𝟏 𝟐
𝟐 𝟖

Let’s find the probability distribution of X :

First methode

𝐺𝑋(𝑡) can be written as follows:

𝟏 𝟏 𝟏 𝟏
𝑮𝑿 𝒕 = 𝟖
𝒕𝟎 + 𝟒
𝒕 + 𝟖
𝒕𝟐 + 𝟐
𝒕𝟑
On the other hand, we know that:

𝑮𝑿 𝒕 = 𝒕𝟎 𝑷 𝑿 = 𝟎 + 𝒕𝟏 𝑷 𝑿 = 𝟏 + 𝒕𝟐 𝑷 𝑿 = 𝟐 + 𝒕𝟑 𝑷 𝑿 = 𝟑 +…
 𝑮𝑿 𝒕 =
𝟏 𝟎
𝒕 +
𝟏
𝒕 +
𝟏 𝟐
𝒕 +
𝟏 𝟑
𝒕 ①
38
𝟖 𝟒 𝟖 𝟐

𝑮𝑿 𝒕 = 𝒕𝟎 𝑷 𝑿 = 𝟎 + 𝒕𝟏 𝑷 𝑿 = 𝟏 + 𝒕𝟐 𝑷 𝑿 = 𝟐 + 𝒕𝟑 𝑷 𝑿 = 𝟑 +… ②

By identification between ① et ② , we find :

𝟏
𝑷 𝑿=𝟎 =
𝟖
𝟏
𝑷 𝑿=𝟏 =
𝟒
𝟏
𝑷 𝑿=𝟐 =
𝟖
𝟏
𝑷 𝑿=𝟑 =
𝟐
Second methode: 39
(𝒌)
𝐆𝐗 𝟎
By using the property: 𝑷 𝑿 = 𝒌 = 𝒌!

𝟏 𝟎 𝟏 𝟏 𝟐 𝟏 𝟑 𝟏
𝑮𝑿 𝒕 = 𝒕 + 𝒕 + 𝒕 + 𝒕 𝑷 𝑿=𝟎 = 𝑮𝑿 𝟎 =
𝟖 𝟒 𝟖 𝟐 𝟖

𝟏
𝟏 𝟏 𝟑 𝑷 𝑿=𝟏 = 𝑮′𝑿 𝟎 =
𝐆𝐗′ 𝒕 = 𝟒
+𝟒 𝒕 +𝟐 𝒕𝟐 𝟒

𝟏 ′′ 𝟏
𝐆𝐗′′ 𝒕 =
𝟏
+𝟑𝒕 𝑷 𝑿=𝟐 = 𝑮𝑿 𝟎 =
𝟒 𝟐 𝟖

(𝟑) 𝟏 (𝟑) 𝟏
𝐆𝐗 𝒕 =𝟑 𝑷 𝑿=𝟑 = 𝑮𝑿 𝟎 =
𝟔 𝟐
Property 2: 40
Let 𝐺𝑋(𝑡) be the Probability Generating Function of a
discrete random variable X defined on 𝑋(Ω),

For t = 1, 𝐺𝑋(𝑡) can be used in order to find the factorial

moments as follows:
(𝒌)
 𝑮𝑿 𝟏 = 𝑬 (𝑿) 𝑿 − 𝟏 … (𝑿 − 𝒌 + 𝟏)

 𝑮𝑿 𝟏 = 𝑬(𝟏𝑿 ) = 𝑬(𝟏)=1

 𝑮′𝑿 𝟏 = 𝑬(𝑿)

 𝑮′′
𝑿 𝟏 = 𝑬 (𝑿) 𝑿 − 𝟏
 41
Let’s take the second order factorial moment:

𝐄 (𝐗) 𝐗 − 𝟏 = 𝐆𝐗′′ 𝟏

⇒ 𝐄 𝐗𝟐 − 𝐗 = 𝐆𝐗′′ 𝟏

⇒ 𝐄 𝐗 𝟐 − 𝐄 𝐗 = 𝐆𝐗′′ 𝟏

Then, we can deduce that:

𝐄 𝐗 𝟐 = 𝐆𝐗′′ 𝟏 + 𝐆𝐗′ 𝟏
Example: 42
By using the PGF of the previous example of the die,

𝟏 𝟐 𝟐 𝟑 𝟑 𝟒 𝟒 𝟓 𝟓 𝟔 𝟔
𝑮𝑿 𝒕 = 𝒕+ 𝒕 + 𝒕 + 𝒕 + 𝒕 + 𝒕
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
We can find the factorial moments and deduce the
expectation and the variance as follows:
The first order factorial moment (Expectation):

𝟏 𝟒 𝟗 𝟐 𝟏𝟔 𝟑 𝟐𝟓 𝟒 𝟑𝟔 𝟓
𝐆𝐗′ 𝒕 = + 𝒕+ 𝒕 + 𝒕 + 𝒕 + 𝒕
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
𝟏 𝟒 𝟗 𝟏𝟔 𝟐𝟓 𝟑𝟔
⇒ 𝐄(𝐗) = 𝐆𝐗′ 𝟏 = + + + + + = 𝟒, 𝟑𝟑
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
The second order factorial moment : 43
𝟒 𝟏𝟖 𝟒𝟖 𝟐 𝟏𝟎𝟎 𝟑 𝟏𝟖𝟎 𝟒
𝐆𝐗′′ 𝒕 = + 𝒕+ 𝒕 + 𝒕 + 𝒕
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏
𝟒 𝟏𝟖 𝟒𝟖 𝟏𝟎𝟎 𝟏𝟖𝟎
𝐄 𝐗 −𝐗 𝟐
= 𝐆𝐗′′ 𝟏 = + + + + = 𝟏𝟔, 𝟔𝟕
𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏 𝟐𝟏

then , we can deduce :

- The second order raw moment :
𝐄 𝐗 𝟐 = 𝐆𝐗′′ 𝟏 + 𝐆𝐗′ 𝟏 = 𝟒, 𝟑𝟑 + 𝟏𝟔, 𝟔𝟕 = 𝟐𝟏
- The variance:
𝑽 𝑿 = 𝑬 𝑿𝟐 − (𝑬 𝑿 )𝟐 = 𝟐𝟏 − 𝟒, 𝟑𝟑 𝟐
= 𝟐, 𝟐𝟐
 44

Property 3:

If 2 random variables X and Y have the same probability

generating function (i.e. G𝑋 (t) = G𝑌(t) ), then: 𝑋 et 𝑌 have
the same probability distribution.
4. Transformations of random variables 45
(Transformation de variables alétoires)

Definition :
Let 𝑋 be a discrete random variable on (Ω,𝒜,𝑃) and let
𝑌=g(𝑋) be a transformation of 𝑋.

𝑌 is a random variable on(Ω,𝒜,𝑃), its PMF is given by :

𝑷(𝒀 = 𝒚𝒋 ) = 𝑷(𝑿 = 𝒙𝒊 )
𝒙𝒊 ∈𝑿 𝛀 :𝐠 𝒙𝒊 =𝒚𝒋
 46

In order to determine the probability distribution of Y,

we have to :
1. Find Y(Ω).
2. Find P(Y = yj ) = P(g(X) = yj ), ∀yj ∈ Y(Ω).
Example: 47
Let 𝑋 be a random variable with the following probability distribution :

xi -3 -2 -1 0 1 2 3
P (X= xi) 0,1 0,15 0,2 0,25 0,05 0,1 0,15

Let y be another random variable such as: 𝑌= 𝑋² + 2

We start by finding 𝑌 (Ω)

For 𝑋= - 3 or 𝑋= 3 ; 𝑌= 11
For 𝑋= - 2 or 𝑋= 2 ; 𝑌= 6
For 𝑋= -1 or 𝑋= 1 ; 𝑌= 3
For 𝑋= 0 ; 𝑌= 2

𝑌 (Ω) = {2, 3, 6, 11}.

 xi -3 -2 -1 0 1 2 3 48
P (X= xi) 0,1 0,15 0,2 0,25 0,05 0,1 0,15

The probability distribution of 𝑌 is given by:

yj 2 3 6 11

P (𝑌 = yj) 0,25 0,25 0,25 0,25

P(𝑌= 2) = P(𝑋= 0) = 0,25 ; 𝑌= 2 ⇔ 𝑋= 0 ;

P(𝑌= 3) = P(𝑋= - 1)+P(𝑋= 1) = 0,25 ; 𝑌= 3 ⇔ 𝑋= -1 or 𝑋= 1 ;
P(𝑌= 6) = P(𝑋= - 2)+P(𝑋= 2) = 0,25 ; 𝑌= 6 ⇔ 𝑋= - 2 or 𝑋= 2 ;
P(𝑌= 11) = P(𝑋= - 3)+P(𝑋= 3) = 0,25 ; 𝑌= 11 ⇔ 𝑋= - 3 or 𝑋= 3 ;
5. Markov and Chebyshev inequalities 49
(Inégalités de Markov et de Chebyshev )

 Markov’s inequality

Theorem:
Let 𝑋 be a non negative random variable defined
on the probability space (Ω,𝒜,𝑃), then:

𝐄(𝐗)
∀𝐚 > 𝟎, 𝐏(𝐗 ≥ 𝐚) ≤
𝒂
 50

 Chebyshev’s inequality

Theorem:
Let 𝑋 be a non-negative random variable defined
on the probability space (Ω,𝒜,𝑃), then:

𝑽(𝑿)
∀𝐚 > 0, 𝐏( 𝐗 − 𝐄(𝐗) ≥ 𝐚) ≤
𝒂2
Proof 51

Since 𝑌=(𝑋−𝐸(X) )2 is a non-negative random variable,

we can apply Markov’s inequality (with 𝑏 =𝑎2):

E[Y]
P(Y ≥ b) ≤ 𝑏

2
2 2
E[ X − E(X) ]
P( X − E(X) ≥ 𝑎 )≤
𝑎2

Then:
𝑉(𝑋)
P( X − E(X) ≥ 𝑎) ≤ 𝑎2
Example 52
Let X be a random variable the associates the number of
customers who come to a store in a day, with E(X) = 50 and
V(X)=36.

* By using Markov’s inequality, the probability that the number of

customers on a given day exceeds 80 is estimated by :

𝟓𝟎
𝐏(𝐗 ≥ 𝟖𝟎) ≤ Then : 𝐏(𝐗 ≥ 𝟖𝟎) ≤ 𝟎, 𝟔𝟐𝟓
𝟖𝟎

* By using Chebyshev’s inequality, this probability is estimated by:

𝟑𝟔
𝐏( 𝐗 − 𝟓𝟎 ≥ 𝟑𝟎) ≤ Then:
𝟑𝟎𝟐

𝐏 𝐗 − 𝟓𝟎 ≥ 𝟑𝟎 ≤ 𝟎, 𝟎𝟒