Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Hanoi University of Science and Technology

Applied Statistics for Business

Introduction

evans.gouno@univ-ubs.fr
2022–2023

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

1 Random variables

2 Distribution Functions

3 Exponential

4 Gamma

5 Normal

6 Central Limit Theorem

7 Distribution for Statistics

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Random variables

Definition
A random variable (r.v.) X is a mapping that assigns a real number to each
outcome:
X : Ω −→ V
ω 7−→ x = X (ω)

A random variable can be

discrete, that is V ⊆ N. (ex: {0, 1}, {1, 2, 3, 4}, N, etc.)

continuous, that is V ⊆ R (ex: [0, 1] ou R, etc.).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Examples of random variables (r.v.)

• discrete:
number of failures of a system component in a given time period,

number of {Head} obtained after flipping a coin n times,

number of dots shown after rolling two six-sided dice.

• continue:
times to failure of a device,

necessary time to perfom a given task,

weights, heights of individuals in a population.

measurement errors.

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Discrete random variables

Random experiment: Tossing a coin.

Possible outcomes: Tail or Head.

Universe: 
Ω = Tail, Head

Associated random variable: X (Tail) = 0 et X (Head) = 1.


V = 0, 1 .

Tail = X −1 (0) and Head = X −1 (1).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Distribution of a discrete r.v.

When a random variable is discrete, its law or distribution is described by

the quantities P(X = x ), x ∈ V .

We have:
X
P(X = x ) ∈ [0, 1], ∀x ∈ V et P(X = x ) = 1.
x ∈V

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Continuous random variables

X : Lap time for a given operation

Ω = R+ .

A possible outcome is: 1h, 27 min., 30 sec., 5/10 msec., . . .

In this case, it is not possible to associate an exact value to the outcome.

Event: X −1 (]a, b[) = ω ∈ Ω | a < X (ω) < b .



The distribution is characterized by:

P X −1 (]a, b[) = P a < X (ω) < b = P a < X < b .

  

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Distribution function

When a random variable is continuous, we consider:

P(x < X 6 x + h) = P(X 6 x + h) − P(X 6 x ).

We define the cumulative distribution function (c.d.f.).

Definition
The cumulative distribution function is the function FX : R −→ [0, 1] defined
by
FX (x ) = P(X 6 x ).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Probability density function

Continuous random variable

We define fX (x ), the probability density function of X as:
1

fX (x ) = lim F (x + h) − F (x ) .
h→0 h

The pdf is the derivative of FX , the cdf:

fX (x ) = F ′ (x ).

We have:
Z Z
FX (x ) = P(X 6 x ) = fX (y )dy and fX (y )dy = 1.
{y 6x } V

Remark: Discrete random variable

X
FX (x ) = P(X 6 x ) = P(X = y ).
{y ∈V |y 6x }

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exponential distribution

Definition
A r.v. X has an exponential distribution with parameter θ > 0 iff its p.d.f. has
the form:
1 x
n o
f (x | θ) = exp − , x ∈ R+ .
θ θ

Notation: X ∼ E (θ)

Cumulative distribution function: F (x ) = 1 − e −x /θ .

Moment generating function:

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exponential distribution

Definition
A r.v. X has an exponential distribution with parameter θ > 0 iff its p.d.f. has
the form:
1 x
n o
f (x | θ) = exp − , x ∈ R+ .
θ θ

Notation: X ∼ E (θ)

Cumulative distribution function: F (x ) = 1 − e −x /θ .

1
Moment generating function: MX (t) = .
1 − θt

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exponential distribution

Definition
A r.v. X has an exponential distribution with parameter θ > 0 iff its p.d.f. has
the form:
1 x
n o
f (x | θ) = exp − , x ∈ R+ .
θ θ

Notation: X ∼ E (θ)

Cumulative distribution function: F (x ) = 1 − e −x /θ .

1
Moment generating function: MX (t) = .
1 − θt

Expectation:

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exponential distribution

Definition
A r.v. X has an exponential distribution with parameter θ > 0 iff its p.d.f. has
the form:
1 x
n o
f (x | θ) = exp − , x ∈ R+ .
θ θ

Notation: X ∼ E (θ)

Cumulative distribution function: F (x ) = 1 − e −x /θ .

1
Moment generating function: MX (t) = .
1 − θt

Expectation: E (X ) = θ.

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exponential distribution

Definition
A r.v. X has an exponential distribution with parameter θ > 0 iff its p.d.f. has
the form:
1 x
n o
f (x | θ) = exp − , x ∈ R+ .
θ θ

Notation: X ∼ E (θ)

Cumulative distribution function: F (x ) = 1 − e −x /θ .

1
Moment generating function: MX (t) = .
1 − θt

Expectation: E (X ) = θ.

Variance:

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exponential distribution

Definition
A r.v. X has an exponential distribution with parameter θ > 0 iff its p.d.f. has
the form:
1 x
n o
f (x | θ) = exp − , x ∈ R+ .
θ θ

Notation: X ∼ E (θ)

Cumulative distribution function: F (x ) = 1 − e −x /θ .

1
Moment generating function: MX (t) = .
1 − θt

Expectation: E (X ) = θ.

Variance: Var (X ) = θ2 .

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Gamma distribution

Definition
A r.v. X has a gamma distribution with parameter α, β > 0 iff its p.d.f. has
the form:
βα
f (x | α, β) = x α−1 e −βx , x ∈ R+
Γ(α)
Z +∞
where Γ(.) is the gamma function defined by: Γ(α) = u α−1 e −u du.
0

Notation: X ∼ G (α, β)
Expectation: E (X ) = α/β.
Variance: Var (X ) = α/β 2 .
 α
β
Moment generating function: MX (t) = .
β−t

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal (Gaussian) distribution

Johann Carl Friedrich Gauss (1777-1855) was

a German mathematician and physicist who
worked in a wide variety of fields in both
mathematics and physics including num-
ber theory, analysis, differential geometry,
geodesy, magnetism, astronomy and optics.
His work has had an immense influence in
many areas.

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal (Gaussian) distribution

Definition
A r.v. X has a normal with parameters (µ, σ 2 ) µ > 0, σ 2 > 0 iff its p.d.f. is:
 
2 1 (x − µ)2
f (x | µ, σ ) = √ exp − , x ∈ R.
2πσ 2σ 2

Notation: X ∼ N (µ, σ 2 ).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal (Gaussian) distribution

Definition
A r.v. X has a normal with parameters (µ, σ 2 ) µ > 0, σ 2 > 0 iff its p.d.f. is:
 
2 1 (x − µ)2
f (x | µ, σ ) = √ exp − , x ∈ R.
2πσ 2σ 2

Notation: X ∼ N (µ, σ 2 ).

Cumulative distribution function:

x  
(t − µ)2
Z
1
F (x ) = √ exp − dt.
−∞ 2πσ 2σ 2

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal (Gaussian) distribution

p.d.f. of a normal r.v for differents parameters values

(x −µ)2
1 −
fX (x ) = √ e 2σ 2
2πσ
µ = 0, σ 2 = 1
2
µ = 2, σ = 5
µ = 4, σ 2 = 2

0 x

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

When µ = 0 and σ 2 = 1, X is said to follow the standard normal distribution.

p.d.f. of the standard normal distribution:
 
1 x2
f (x | 0, 1) = φ(x ) = √ exp − , x ∈ R.
2π 2

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

When µ = 0 and σ 2 = 1, X is said to follow the standard normal distribution.

p.d.f. of the standard normal distribution:
 
1 x2
f (x | 0, 1) = φ(x ) = √ exp − , x ∈ R.
2π 2

c.d.f. of the standard normal distribution:

Z x  
1 t2
Φ(x ) = √ exp − dt.
−∞ 2π 2

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

When µ = 0 and σ 2 = 1, X is said to follow the standard normal distribution.

p.d.f. of the standard normal distribution:
 
1 x2
f (x | 0, 1) = φ(x ) = √ exp − , x ∈ R.
2π 2

c.d.f. of the standard normal distribution:

Z x  
1 t2
Φ(x ) = √ exp − dt.
−∞ 2π 2

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Proposition
X −µ
 
X ∼ N (µ, σ 2 ) ⇐⇒ Z =


∼ N (0, 1) .
σ

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Expectation
Let Z ∼ N (0, 1), then E (Z ) = 0.
Let X = µ + σZ , X ∼ N (µ, σ 2 ) and E (X ) = µ.
Variance
Var (Z ) = E (Z 2 ) − (EZ )2 .
It can be shown that: E (Z 2 ) = 1. Thus: Var (Z ) = 1.
Var (X ) = Var (µ + σZ ) = σ 2 Var (Z ) = σ 2 .
Conclusion
If X ∼ N (µ, σ 2 ), then E (X ) = µ and Var (X ) = σ 2 .

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Proposition

∀z, Φ(z) + Φ(−z) = 1.

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Proposition

∀z, Φ(z) + Φ(−z) = 1.

1 −x 2 /2
φ(x ) = √ e
2π

P(Z 6 −z) = Φ(−z) P(Z > z) = 1 − P(Z 6 z) = 1 − Φ(z)

−z 0 z x

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Proposition
Let Z ∼ N (0, 1), P(|Z | 6 a) = 2Φ(a) − 1.

1 −x 2 /2
φ(x ) = √ e
2π

P(|Z | 6 a|) = P(−a < Z 6 a)

= P(Z 6 a) − P(Z 6 −a)
= Φ(a) − Φ(−a)
= 2Φ(a) − 1

−a a x

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Remark: P(|Z | > a) = 1 − P(|Z | 6 a) = 1 − 2Φ(a) + 1 = 2(1 − Φ(a)).

1+γ
Let P(|Z | 6 a) = γ, then Φ(a) = .
2
1+γ
 
The quantile a associated with the probability γ is a = Φ−1 .
2

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Normal distribution

Let consider X ∼ N (µ, σ 2 ).

We have:
X −µ
 
P 6 a = γ ⇐⇒ P(µ − aσ < X 6 µ + aσ) = γ.
σ
Therefore the probability that the r.v. lies in an interval of length 2aσ centered
on µ is equal to γ.
For a = 1, Φ(1) = 0.8413 and γ = 2 × 0.8413 − 1 = 0.6826.
For a = 2, Φ(2) = 0.9772 and γ = 2 × 0.9772 − 1 = 0.9544.
For a = 3, Φ(3) = 0.9986 and γ = 2 × 0.9986 − 1 = 0.9973.

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exercises
I. Let X be the cost of natural gas per metric cubic foot (MCF) for customers.
Suppose that X is normally distributed with a mean of $ 6.00 and a standard
deviation of $ 1.20.
1 What is the probability that the cost of natural gas per MCF for a
particular customer is between $ 7.60 and $ 8.00?
2 What is the median cost per MCF for natural gas?
3 What are the upper and lower quartiles for the cost per MCF of natural
gas?
II. Let X be the length of life of a type of automatic washer.
Suppose that this random variable is approximately normally distributed with
mean 3.1 years and standard deviation of 1.2 years.
If this type of washer is guaranteed for one year, what fraction of original sales
will require replacement?

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

The statistic X n

Central Limit Theorem (CLT)

Let X1 , · · · , Xn be n independent and identically distributed (iid) r.v.’s with
E (X1 ) = µ and Var (X1 ) = σ 2 > 0 (both finite).
Then:  
√ Xn − µ
n ∼ N (0, 1) lorsque n → +∞.
σ
 
√ Xn − µ
We say that the r.v. n convergence in distribution (or in law) to a
σ
standard normal
 r.v. 
d L
Notation: Zn → Z or Zn → Z ⇐⇒ lim FZn (z) = FZ (z).
n→+∞

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

CLT

Example: Let X1 , · · · , Xn be independent Bernoulli r.v.’s with parameter p. We

have E (Xi ) = p and Var (Xi ) = p(1 − p).
Applying the CLT leads to :
!
√ Xn − p
n p ∼ N (0, 1).
p(1 − p)

Simulation:
Choose a value for p e.g. p = 0.8.
Draw 100 realizations of a Bernoulli with parameter 0.8.
Compute X n , the mean of the 100 values.
Reproduce this experiment 1000 times to obtain a sample of X n .
build the histogram to investigate the distribution of the variables
√ p

n (X n − p)/ p(1 − p) .

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Illustration of the CLT

CLT for Bernoulli

Density

−3 −2 −1 0 1 2 3

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

R code

p<-0.8 # Choose a value for p

n<-100 # Size of the sample
# Generate a sample of n Bernoulli r.v.’s realizations
x<-sample(c(0,1), n, replace=TRUE, prob=c(0.2,0.8))
M<-1000 # Number of psamples
√
Let Z be n(X n − p)/ p(1 − p)
Z<-rep(0,M) # Initialization
for (i in 1:M)
Z[i]<-sqrt(n)*((mean(rbinom(n,1,p))-p)/sqrt(p*(1-p)))
# Histogram
hist(Z, nclass=20, probability=T, main=list("CLT for Bernoulli",cex=1.5),
ylab="Density",yaxt = "n")
# Adding the p.d.f.
z<-seq(-3,3,0.01)
gaussz<-dnorm(z,0,1)
lines(z,gaussz)
box() # To add a frame

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exercise:

Illustrate the CLT with X following a Poisson distribution.

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Exercise:

Illustrate the CLT with X following a Poisson distribution.

lambda<-3.45
M<-1000
Z<-rep(0,M)
for (i in 1:M)
{
x<-rpois(n,lambda)
Z[i]<-sqrt(n)*((sum(x)/n-lambda)/sqrt(lambda))
}
Z
hist(Z,nclass=20,probability=TRUE, col="yellow", main="CLT for Poisson")
y<-sort(Z)
lines(y,dnorm(y), col="green", lwd="2")

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

Chi-square distribution

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

Chi-square distribution
The summation of the square of n independent standard normal r.v.’s follows a
Chi-square distribution with n degrees of freedom (d.o.f.).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

Chi-square distribution
The summation of the square of n independent standard normal r.v.’s follows a
Chi-square distribution with n degrees of freedom (d.o.f.).

Definition
A r.v. X has a chi-square distribution with n degrees of freedom (d.o.f.) iff its
p.d.f has the form:

2−n/2 n/2−1 −x /2
fX (x ) = x e , x ∈ R+ , n ∈ N∗ .
Γ(n/2)

Notation: X ∼ χ2 (n). The c.d.f. is denoted: χ2n (x ) = P(X 6 x ).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

Chi-square distribution
The summation of the square of n independent standard normal r.v.’s follows a
Chi-square distribution with n degrees of freedom (d.o.f.).

Definition
A r.v. X has a chi-square distribution with n degrees of freedom (d.o.f.) iff its
p.d.f has the form:

2−n/2 n/2−1 −x /2
fX (x ) = x e , x ∈ R+ , n ∈ N∗ .
Γ(n/2)

Notation: X ∼ χ2 (n). The c.d.f. is denoted: χ2n (x ) = P(X 6 x ).

The p.d.f. of a Chi-square r.v. with n d.o.f. is the p.d.f of a gamma

distribution with parameters (n/2, 1/2).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Chi-square distribution

P.d.f. of a Chi-2 r.v for different d.o.f.

0.5 n=2
n=5
0.4

n=10
0.3
Density

0.2
0.1
0.0

0 5 10 15

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

The Student’s t-distribution

Let Z ∼ N (0, 1) and Y ∼ χ2 (n) be two indepedent r.v.’s, then:
Z
T = p
Y /n
follows a Student’s t-distribution with n degrees of freedom.

Notation: Z ∼ T (n).

The p.d.f. of a Student’s t-distribution is:

 −(n+1)/2
Γ[(n + 1)/2] t2
fT (t) = √ · 1+ , t ∈ R+ .
n π Γ(n/2) n

The c.d.f. is denoted: tn (x ).

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

The Student’s t-distribution

P.d.f. of the Student’s distribution for n1 = 8 and n2 = 20, 8, 2

0.4
n=1
n=5

0.3
n=10
Density

0.2
0.1
0.0

−6 −4 −2 0 2 4 6

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

Combinations of normal r.v.’s

Fisher distribution
Let Y1 ∼ χ2 (n1 ) and Y2 ∼ χ2 (n2 ), be two independent r.v.’s, then
Y1 /n1
Y =
Y2 /n2
follows a Fisher distribution with (n1 , n2 ) degrees of freedom.

Notation: Y ∼ F (n1 , n2 ).

The p.d.f. of a Fisher distribution with (n1 , n2 ) degrees of freedom is:

n1 /2
Γ[(n1 + n2 )/2] n1 y (n1 −2)/2

fY (y ) = · · .
Γ(n1 /2) Γ(n2 /2) n2 (1 + n1 y /n2 )(n1 +n2 )/2

Introduction Applied Statistics for Business

Random variables Distribution Functions Exponential Gamma Normal Central Limit Theorem Distribution for Statistics

The Fisher distribution

P.d.f. of the Fisher distribution for n1 = 8 and n2 = 20, 8, 2

0.8
n2=20
n2=8
n2=2
0.6
Density

0.4
0.2
0.0

0 1 2 3 4 5 6

Introduction Applied Statistics for Business