Chapter 3: Some special distributions

Phan Thi Khanh Van

E-mail: khanhvanphan@[Link]

February 19, 2025

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 1 / 43
Table of Contents

1 Bernoulli distribution

2 Binomial distribution

3 Poisson distributions. Poisson processes

4 Exponential distributions

5 Uniform distributions.

6 Normal distributions. Central limit theorems

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 2 / 43
Bernoulli distribution

1 Flip a coin. X is the number of heads. P(X = 1) = 0.5= P(X = 0).

2 A worn machine tool produces 1% defective parts. Let X = number of defective
part produced in one trial. P(X = 1) = 0.01, P(X = 0) = 0.99
3 Each sample of air has a 10% chance of containing a particular rare molecule. Let
X = the number of air samples that contain the rare molecule when the next sample
is analyzed. P(X = 1) = 0.1.
4 Of all bits transmitted through a digital transmission channel, 10% are received in
error. Let X = the number of bits in error when the next bit is transmitted.
P(X = 1) = 0.1, P(X = 0) = 0.9.
5 Of all patients suffering a particular illness, 35% experience improvement from a
particular medication. A patient is administered the medication, let X = the number
of patients who experience improvement. P(X = 1) = 0.35, P(X = 0) = 0.75.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 3 / 43
Bernoulli distribution
The random variable X is called a Bernoulli random variable if it takes the value 1 with
probability p and the value 0 with probability q = 1 − p.

Mean and Variance

If X is Bernoulli random variable with parameter p,
µ = E(X ) = p, σ 2 = V(X ) = p(1 − p).

Example
A worn machine tool produces 1% defective parts. Let X = number of defective part
produced in one trial. P(X = 1) = 0.01, P(X = 0) = 0.99.

E(X ) = 0.01
V(X ) = 0.01 × 0.99 = 0.0099.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 4 / 43
Binomial distribution

1 Flip a coin 3 times. Let X = number of heads obtained.

2 A worn machine tool produces 1% defective parts. Let X = number of defective
parts in the next 25 parts produced.
3 Each sample of air has a 10% chance of containing a particular rare molecule. Let
X = the number of air samples that contain the rare molecule in the next 18
samples analyzed.
4 Of all bits transmitted through a digital transmission channel, 10% are received in
error. Let X = the number of bits in error in the next 5 bits transmitted.
5 A multiple-choice test contains 10 questions, each with four choices, and you guess
at each question. Let X = the number of questions answered correctly.
6 In the next 20 births at a hospital, let X = the number of female births.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 5 / 43
Binomial distribution

Flip 3 coins
Flip 3 coins, consider the number of heads:
x 0 1 2 3
1 3 3 1
P(X = x) 8 8 8 8

Camera flash tests

The time to recharge the flash is tested in three cell-phone cameras. The probability that
a camera passes the test is 0.8, and the cameras perform independently. Consider the
number of cameras that pass the test.
x 0 1 2 3
P(X = x) 0.23 C31 × 0.2 × 0.82 C32 × 0.22 × 0.8 0.83

Digital Channel
The chance that a bit transmitted through a digital transmission channel is received in
error is 0.1. Also, assume that the transmission trials are independent. Let X = the
number of bits in error in the next four bits transmitted.
x 0 1 2 3 4
P(X = x) 0.94 C41 × 0.1 × 0.93 C42 × 0.12 × 0.92 C43 × 0.13 × 0.9 0.14

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 6 / 43
Binomial distribution.
A random experiment consists of n Bernoulli trials such that
1 The trials are independent.
2 Each trial results in only two possible outcomes, labeled as “success” and “failure.”
3 The probability of a success in each trial, denoted as p, remains constant.
The random variable X that equals the number of trials that result in a success is a
binomial random variable with parameters 0 < p < 1 and n = 1, 2, · · · .
The probability mass function (PMF) of X is
f (x) = Cnx p x (1 − p)n−x , x = 0, 1, , · · · , n

Mean and Variance

If X is a binomial random variable with parameter p,
µ = E(X ) = np, σ 2 = V(X ) = np(1 − p).

Digital Channel
The chance that a bit transmitted through a digital transmission channel is received in
error is 0.1. Let X = the number of bits in error in the next four bits transmitted
µ = np = 4 × 0.1 = 0.4.
σ 2 = np(1 − p) = 4 × 0.1 × 0.9 = 0.36
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 7 / 43
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 8 / 43
Example
A multiple-choice test contains 25 questions, each with four answers. Assume that a
student just guesses on each question. Find the probability
(a) that the student answers more than 20 questions correctly?
(b) that the student answers fewer than 5 questions correctly?

Let X be the number of questions that the student answered correctly. X is a binomial
random variable with the parameter p = 0.25.
(a) The probability that the student answers more than 20:
P(X > 20)= P(X = 21) + P(X = 22) + P(X = 23) + P(X = 24) + P(X = 25)
21
= C25 .0.2521 0.754 +C25
22
.0.2522 0.753 +C25
23
.0.2523 0.752 +C25
24
.0.2524 0.751 +C25
25
.0.2525
= 9.6769.10−10 .
(b) The probability that the student answers fewer than 5:
P(X < 5)= P(X = 0) + P(X = 1) + P(X = 2) + P(X = 3) + P(X = 4)
0
= C25 .0.7525 +C25
1
.0.251 0.7524 +C25
2
.0.252 0.7523 +C25
3
.0.253 0.7522 +C25
4
.0.254 0.7521
= 0.2137.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 9 / 43
Random processes

In real-life applications, we are often interested in multiple observations of random values

over a period of time. For example S(t) - stock price at the time t ∈ [0, ∞). At any
fixed time t∗ , S(t∗ ) is a random variable. S(t), t ∈ [0, ∞) - random process or
stochastic process.

Random processes
A random process is a collection of random variables usually indexed by time.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 10 / 43
Continuous-time and discrete-time random processes
A continuous-time random process is a random process {X (t), t ∈ J}, where J is
an interval on the real line such as [−1, 1], [0, ∞), (−∞, ∞), etc.
A discrete-time random process (or a random sequence) is a random process
{X (n) = Xn , n ∈ J}, where J is a countable set such as N or Z.

Let N(t) ∈ {0, 1, 2...} be the number of customers who have visited a bank from
t = 9 am until time t, on a given day, for t ∈ [9, 16]- continuous-time.

Let W (t) ∈ R be the thermal noise voltage generated across a resistor in an electric
circuit at time t, for t ∈ [0, ∞) - continuous-time.

Let X (tn ) = Xn ∈ [25, 39] be the temperature in HCM city in the n − th day of May,
n = 1, 2, 3....31 - discrete-time.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 11 / 43
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 12 / 43
Poisson processes

The Poisson process is one of the most widely-used counting processes. It is usually used
in scenarios where we are counting the occurrences of certain events that appear to
happen at a certain rate, but completely at random.
For example:

earthquakes occur in a certain area with a rate of λ = 2 per month and the timings
of earthquakes seem to be completely random.

the number of car accidents at a site or in an area;

the location of users in a wireless network;

the requests for individual documents on a web server;

the outbreak of wars;

photons landing on a photodiode.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 13 / 43
Poisson distribution

Poisson random variable

The random variable X that equals the number of events that occur in the interval of
length T in a Poisson process with the average rate λ > 0 (events/unit) is a Poisson
random variable with parameter 0 < λ. Notation: Pois(λT )
e −λT (λT )x
f (x) = , x = 0, 1, 2...
x!

Mean and Variance

If X is a Poisson random variable over an interval of length T with parameter λ, then
µ = E(X ) = λT , σ 2 = V(X ) = λT .

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 14 / 43
Poisson distribution

Example 1
The number of telephone calls that arrive at a phone exchange is often modeled as a
Poisson random variable. Assume that on the average there are 10 calls per hour.
a) What is the probability that there are exactly 5 calls in one hour?
b) What is the probability that there are 3 or fewer calls in one hour?
c) What is the probability that there are exactly 15 calls in two hours?
d) What is the probability that there are exactly 5 calls in 30 minutes?

Let X , Y , Z be the number of calls that arrive at a phone exchange in 1, 2, 0.5 hours,
respectively.
e −10 105
a) λ = 10, T = 1: P(X = 5) = 5!
= 0.0378.
b) P(X ≤ 3) = P(X = 0) + P(X = 1) + P(X = 2) + P(X = 3)=
 0 1 2 3

e −10 100! + 101! + 102! + 103! = 0.0103.
e −20 2015
c) T = 2, P(Y = 15) = 15!
≈ 0.0516.
e −5 55
d) T = 0.5, P(Z = 5) = 5!
≈ 0.1755.
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 15 / 43
Example 2
Astronomers treat the number of stars in a given volume of space as a Poisson random
variable. The density in the Milky Way Galaxy in the vicinity of our solar system is one
star per 16 cubic light-years.
a) What is the probability of two or more stars in 16 cubic light-years?
b) How many cubic light-years of space must be studied so that the probability of one
or more stars exceeds 0.95?

a) Let X be the number of stars in 16 cubic light-years. λ = 1.

In 16 cubic light-years: T = 1: P(X ≥ 2)
e −1 10 e −1
1 − P(X = 0) − P(X = 1)= 1 − 0!
− 1!
= 0.2642.
b) Let N be the number of cubic light-years must be studied and X be the number of
N
stars in N cubic light-years. We have: T = 16
.
We must have P(Y ≥ 1) > 0.95 ⇔ 1 − P(Y = 0) > 0.95
−N
⇔ P(Y = 0) = e −1.T = e 16 < 0.05
⇔ N > −16 ln 0.05 = 47.9317 (cubic light-years).

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 16 / 43
Example 3
The number of customers arriving at a grocery store can be modeled by a Poisson
process with intensity λ = 10 customers per hour.
a) Find the probability that there are 2 customers between 10 : 00 and 10 : 20.
b) Find the average number of customers who arrive between 10 : 00 and 10 : 20.
c) Find the probability that there are 3 customers between 10 : 00 and 10 : 20 and 7
customers between 10 : 20 and 11 : 00.

λ = 10, The interval I1 : 10 : 00 − 10 : 20, I2 : 10 : 20 − 11 : 00. The length of these

interval T1 = 31 , T2 = 2
3
hours. Let X and Y be the number of customers arriving in the
intervals I1 and I2 , respectively. X , Y are a Poisson random variables.
−10
e 3 ( 10
3
)2
a) Therefore, P(X = 2) = 2!
= 0.1982
1 10
b) E(X ) = λT1 = 10 × 3
= 3
.
c) Because I1 and I2 are two disjoint intervals, we have:
P(3 arrive in I1 , 7 arrive in I2 ) = P(3 arrive in I1 ).P(7 arrive in I2 )
−10 −20
e 3 ( 10
3
)3 e 3 ( 20
3
)7
= P(X = 3).P(Y = 7)= 3!
. 7!
= 0.0325.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 17 / 43
Sum of independent Poisson random variables
Let X1 and X2 be two independent Poisson random variables:
X1 ∼ Pois(λ1 T1 ), X2 ∼ Pois(λ2 T2 ). Then, the sum of them: X = X1 + X2 is also a
Poisson random variable and
X ∼ Pois(λ1 T1 + λ2 T2 )
In general, if X1 ∼ Pois(µ1 ), X2 ∼ Pois(µ2 ), ..., Xn ∼ Pois(µn ) are independent, then:
n
P n
P
X = Xi ∼ Pois( µi )
i=1 i=1

Example
The number of customers arriving at a grocery store can be modeled by a Poisson
process with intensity λ = 10 customers per hour. Let X and Y be the numbers of
customers that arrive between 10 : 00 − 10 : 10 and between 10 : 10 − 10 : 30,
respectively. X ∼ Pois(λT1 = 35 ), Y ∼ Pois(λT2 = 103
).
The total number of customers that arrive between 10 : 00 − 10 : 30: X + Y ∼ Pois(5)

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 18 / 43
Poisson Approximation to the Binomial Distribution

Theorem
Let X ∼ B(n, p), we have: E(X ) = np. If p → 0 and np → µ when n → ∞ then
µx
lim P(X = x) = e −µ .
n→∞ x!
When p is small and n is large enough, we can approximate X by the Poisson random
variable Poisson(np).

Example
In a company, the switchboard give services for 100 telephones. The probability that in 1
minute there is a call from a telephone to the switchboard is p = 0.02. Using the direct
and Poisson approximation methods, find the probability that in one minute there are
three telephones calling the switchboard.

Let X is the number of telephones that call the switch board in 100 telephones. We
have: X ∼ B(100, 0.02).
3
Using the direct method: P(X = 3) = C100 0.023 .0.9897 = 0.1823.
3
Using Poisson approximation: X ≈ Poisson(np = 2): P(X = 3) = e −2 . 23! = 0.1804.
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 19 / 43
Exponential Distribution
The random variable X that equals the distance between successive events from a
Poisson process with mean number of events λ > 0 per unit interval is an exponential
random variable with parameter λ. The PDF of X is
−λx
PDF: f (x) = λe
Rx , for 0 ≤ x < ∞
CDF: F (x) = 0 f (t)dt = 1 − e −λx .

Mean and variance

If the random variable X has exponential distribution with parameter λ,
µ = E(X ) = 1
λ
, and σ 2 = V(X ) = 1
λ2

Memoryless
If the random variable X has exponential distribution with parameter λ, then X is
memoryless random variable, that is
P(X > x + a X > a) = P(X > x), for a, x ≥ 0

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 20 / 43
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 21 / 43
Computer Usage
In a large corporate computer network, user log-ons to the system can be modeled as a
Poisson process with a mean of 25 log-ons per hour. What is the probability that there
are no log-ons in an interval of six minutes?

Let X denote the time in hours from the start of the interval until the first log-on. Then
X has an exponential distribution with λ = 25 log-ons per hour.
We are interested in the probability that X exceeds 6 minutes. Because λ is given in
log-ons per hour, we express all time units in hours. That is, 6 minutes = 0.1 hour.
R∞
Therefore, P(X > 0.1) = 25e −25x dx = e −25(0.1) = 0.082.
0.1
The CDF also can be used to obtain the same result as follows:
P(X > 0.1) = 1 − F (0.1) = e −25(0.1) = e −2.5 .

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 22 / 43
Example 2
Web crawlers need to estimate the frequency of changes to Web sites to maintain a
current index for Web searches. Assume that the changes to a Web site follow a Poisson
process with a mean of 3.5 days.
a) What is the probability that the next change occurs in less than 2.0 days?
b) What is the probability that the time until the next change is greater 7.0 days?
c) What is the time of the next change that is exceeded with probability 90%
d) What is the probability that the next change occurs in less than 10.0 days, given
that it has not yet occurred after 3.0 days?

Let X denote the time in days from 2 changes of Web sites. X is an exponential random
x Rx 1 − t −x
1 1 − 3.5
variable with λ = 3.5 . PDF: f (x) = 3.5 e . CDF: F (x) = 0 3.5 e 3.5 dt= 1 − e 3.5
a) P(X < 2) = F (2)= 0.4353.
b) P(X > 7) = 1 − P(X < 7) = 1 − F (7)= 0.1353
−x0
c) P(X > x0 ) = 1 − P(X < x0 ) = e 3.5 = 0.9⇒ x0 = −3.5 ln 0.9 = 0.3688
d) Using the memoryless property of the exponential random variable:
7
P(X < 10|X > 3) = P(X < (10 − 3)) = P(X < 7) = F (7) = 1 − e − 3.5 = 0.8647.
F (10) − F (3)
Or we can use: P(X < 10|X > 3) = = 0.8647.
1 − F (3)
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 23 / 43
Uniform distributions.

Uniform distributions
A continuous(uniform random variable X is a continuous random variable with:
1
, a ≤ x ≤ b,
PDF: f (x) = b−a .
0, x < a or x > b

0
 x < a,
CDF: F (x) = x−a
b−a
, a ≤ x ≤ b,

1, x >b


Mean and Variance

If X is a continuous uniform random variable over a ≤ x ≤ b
(a+b) (b−a)2
µ = E(x) = 2
, and σ 2 = V(X ) = 12
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 24 / 43
Example 1
An e-mail message will arrive at a time uniformly distributed between 9 : 00 a.m. and
11 : 00 a.m. You check e-mail at 9 : 15 a.m. and every 30 minutes afterward.
a) What is the standard deviation of arrival time (in minutes)?
b) What is the probability that the message arrives less than 10 minutes before you
view it?
c) What is the probability that the message arrives more than 15 minutes before you
view it?

Let X be the time that an email message arrive.

q q
2
σ = V(X ) = (11−9)
p
a)
12
= 13 (hour).
1
b) PDF: f (x) = b−a
= 0.5 The probability that the message arrives less than 10
minutes before you view it:
P(9 : 05 ≤ X ≤ 9 : 15) + P(9 : 35 ≤ X ≤ 9 : 45)
+P(10 : 05 ≤ X ≤ 10 : 15) + P(10 : 35 ≤ X ≤ 10 : 45)= 4 · 0.5 · 61 ) = 0.3333
c) The probability that the message arrives more than 15 minutes before you view it:
P(9 : 15 ≤ X ≤ 9 : 30) + P(9 : 45 ≤ X ≤ 10 : 00) + P(10 : 15 ≤ X ≤ 10 : 30)=
1
0.5 · 3 · 4
= 0.375
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 25 / 43
Example 2
The volume of a shampoo filled into a container is uniformly distributed between 374 and
380 milliliters.
a) What are the mean and standard deviation of the volume of shampoo?
b) What is the probability that the container is filled with less than the advertised
target of 375 milliliters?
c) What is the volume of shampoo that is exceeded by 95% of the containers?
d) Every milliliter of shampoo costs the producer $0.002. Any shampoo more than 375
milliliters in the container is an extra cost to the producer. What is the mean extra
cost (only consider the containers which cause extra cost)?

a) Let X be the volume of a shampoo filled into a container. X is a uniform random

variable with a = 374, b = 380.
E(X ) = a+b
2
= 377
p b−a
√
σ = V(x) = √ 12
= 3.
R 375
b) P(X < 375) = 374 dt6 = 16 .

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 26 / 43
 R 380 dt
c) P(X > x0 ) = 95%⇔ x0 6
= 0.95
380−x0
⇔ 6
= 0.95⇔ x0 = 380 − 6 · 0.95 = 374.3.

d) Let Y be the extra cost for one volume container.

(
0.002 × (X − 375), if 380 ≥ X > 375
Y =
0, if X ≤ 375 ∨ X > 380
The mean extra cost:
R∞ R 380
C = E(Y (X )) = Y (x)f (x)dx = 375 0.002 × (x − 375) × 16 dx = 0.0042
−∞

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 27 / 43
Normal distributions
A random variable X with probability density function
−(x−µ)2
f (x) = √1 e 2σ 2 , −∞ < x < ∞
2πσ

is called a normal random variable with parameters µ where −∞ < µ < ∞, and σ > 0.
E(X ) = µ, V(X ) = σ 2 .
Notation: X ∼ N(µ, σ 2 )

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 28 / 43
Normal distributions

Example 1
Assume that the current measurements in a strip of wire follow a normal distribution
with a mean of 10 milliamperes and a variance of 4 (milliamperes) . What is the
probability that a measurement exceeds 13 milliamperes?

Let X denote the current in milliamperes.

R∞ 1 −(x−10)2
P(X > 13) = √2π.2 e 2.4 dx ≈ 0.0668
13

Standard normal random variable

A normal random variable X with µ = 0, σ 2 = 1 is called a standard normal random
variable and is denoted as Z ∼ N(0, 1). CDF: Φ(Z ) = P(Z ≤ z).
Upper percentile zα is the point such that: P(Z > zα ) = α.

X −µ
If X ∼ N(µ, σ 2 ), using the change the random variable: Z = σ

E (Z ) = 0, V (Z ) = 1 : Z ∼ N(0, 1) : standard normal random variable.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 29 / 43
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 30 / 43
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 31 / 43
Normal distributions

Example 2
The time until recharge for a battery in a laptop computer under common conditions is
normally distributed with a mean of 260 minutes and a standard deviation of 50 minutes.

a) What is the probability that a battery lasts more than four hours?
b) What value of life in minutes is exceeded with 75% probability?

Let X be the time until recharge for a battery in a laptop computer under common
conditions. X ∼ N(260, 502 )
X −µ
a) Change the variable: Z = = X −260
σ50
, we have Z ∼ N(0, 1)
R∞ 1 −z 2
P(X ≥ 240) = P(Z ≥ −0.4)= √ e 2 dz = 0.6554
2π
−0.4
Or we can use the table: P(Z ≥ −0.4) = 1 − P(Z < −0.4) = 1 − 0.3446.
b) Let X0 be the value of life in minutes which is exceeded with 75% probability
X0 −260
P(X > X0 ) = P(Z > zα = 50
) = 1 − Φ(zα ) = 0.75 (= α).
Use the table: zα = −0.67. : X0 = 260 − 50 × 0.67 = 226 (min).
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 32 / 43
Example 3
The weight of a sophisticated running shoe is normally distributed with a mean of 12
ounces and a standard deviation of 0.5 ounce.
a) What is the probability that a shoe weighs more than 13 ounces?
b) What value of weight is exceeded with 95% probability?
c) What must the standard deviation of weight be in order for the company to state that
99.9% of its shoes weighs less than 13 ounces?

Let X be the weight of a sophisticated running shoe.

σ = 0.5, µ = 12.
X −12
Change the variable: Z = 0.5
we have: Z ∼ N(0, 1).
a) X = 13 ⇒ Z = 2.
P(Z > 2) = 1 − P(Z ≤ 2) (use table) = 1 − 0.9773 = 0.0227
b) P(Z > z0 ) = 0.95 ⇒ P(Z ≤ z0 ) = 0.05 (use table) ⇒ z0 = −1.64
⇒ x0 = z0 · 0.5 + 12 = 11.18
c) In order for the company to state that 99.9% of its shoes weighs less than X = 13
ounces: (use table) Z = 3.1
X −µ
⇒σ= Z
= 13−12
3.1
= 0.3226.
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 33 / 43
Sum of independent normal random variables
Consider 2 independent normal random variables X1 ∼ N(µ1 , σ12 ), X2 ∼ N(µ2 , σ22 ). Then,
X = X1 + X2 is also a normal random variable:
X ∼ N(µ1 + µ2 , σ12 + σ22 ).
In general, if we consider n independent normal random variables:
X1 ∼ N(µ1 , σ12 ), X2 ∼ N(µ2 , σ22 ),· · · , Xn ∼ N(µn , σn2 ),
we have
n n n
ci2 σi2 ).
P P P
X = ci Xi ∼ N( c i µi ,
i=1 i=1 i=1

Central limit theorem (CLT)

Let X1 , X2 , · · · be a sequence of independent and identically distributed (i.i.d.) random
variables, each having mean µ and variance σ 2 .
X1 + ....Xn − nµ
Then the distribution of Zn = √ tends to the standard normal N(0, 1) as
σ n
n → ∞. That is, for −∞ < a < ∞,
n o Ra −x 2 /2
√ n −nµ ≤ a → √1
P X1 +···+X e dx, as n → ∞
σ n 2π
−∞

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 34 / 43
Central limit theorem

The condition to use CLT: n ≥ 30 or n ≥ 40.

P
Xi − nµ
Z = √ ≈ N(0, 1), or
σ n P
Xi σ2
Xi ≈ N(nµ, nσ 2 ),
P
X = X̄ = ≈ N(µ, n
).
n

Example
Assume that the average age of a university student is 22.3 (years old) with the standard
deviation 4. Randomly choose 64 students, find the probability that the average age of
these students is greater than 23.

Let Xi , i = 1, 64 be the ages of these 64 students.

We have that Xi are independent random variable of the same distribution:
µ = 22.3, σ = 4.
X1 + · · · + Xn
Consider the average age: X̄ = where n = 64.
n
X̄ − µ
Using CLT: Z = √ ≈ N(0, 1).
σ/ n
We have: P(X̄ > 23) = P(Z > 23−22.3
√
4/ 64
)≈ 1 − Φ(1.4) = 1 − 0.9192 = 0.0808.
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 35 / 43
Central limit theorem

Example
On the way to the university everyday, a student will meet 4 traffic lights. Let Xi be the
number of green lights that he meets in the i − th day. The distribution of Xi is given in
x 0 1 2 3 4
the table
f (x) 0.15 0.2 0.25 0.1 0.3
a. Find E(Xi ), V(Xi ).
b. Find the probability that in 100 days the student meets more than 190 green lights.

P
a. µ = xi f (xi ) = 2.2.
2 P 2
σ = xi f (xi ) − µ2 = 2.06.
b. Consider the total number of green lights in 100 days: X = X1 + X2 + ... + X100 .
Using CLT: X ≈ N(nµ, nσ 2 ) = N(220, 206).
1 R∞ −(x−220)2
We have: P(X > 190) = √ e 2×206 dx ≈ 0.9817.
2π × 206 190

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 36 / 43
Normal Approximation to the Binomial Distribution

Example
Assume that in a digital communication channel, the number of bits received in error can
be modeled by a binomial random variable, and assume that the probability that a bit is
received in error is 10−5 . If 16 million bits are transmitted, what is the probability that
150 or fewer errors occur?

Let the random variable X denote the number of errors. Then X is a binomial random
variable and
150
x
(10−5 )x (1 − 10−5 )16,000,000−x
P
P(X ≤ 150) = C16,000,000
x=0

- this probability is difficult to compute.

We can consider X as the sum of n = 16 million random variables Xi , where
Xi ∼ B(1, 10−5 ) - independent Bernoulli random variables.
E (Xi ) = p, V(Xi ) = p(1 − p)
The normal distribution can be used to provide an excellent approximation in this
example.
(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 37 / 43
Normal Approximation to the Binomial Distribution

Normal Approximation to the Binomial Distribution

If X is a binomial random variable with parameters n and p,
Z = √X −np
np(1−p)

is approximately a standard normal random variable. To approximate a binomial

probability with a normal distribution, a continuity correction is applied as follows:
 
x+0.5−np
P(X ≤ x) = P(X ≤ x + 0.5) ≈ P Z ≤ √
np(1−p)

and
 
x−0.5−np
P(X ≥ x) = P(X ≥ x − 0.5) ≈ P Z ≥ √
np(1−p)

The approximation is good for np > 5 and n(1 − p) > 5.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 38 / 43
Normal Approximation to the Binomial Distribution

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 39 / 43
Return to the previous example
Assume that in a digital communication channel, the number of bits received in error can
be modeled by a binomial random variable, and assume that the probability that a bit is
received in error is 10−5 . If 16 million bits are transmitted, what is the probability that
150 or fewer errors occur?

Let the random variable X denote the number of errors. Then X is a binomial random
variable: X ∼ B(np, np(1 − p)).
 
X −160
P(X ≤ 150) = P(X ≤ 150.5) = P √ ≤ √150.5−160−5 ≈ P (Z ≤ −0.75) =
160(1−10−5 ) 160(1−10 )
0.227.

Here, np = 16, 000, 000 · 10−5 = 160, n(1 − p) > 5.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 40 / 43
Normal Approximation to Poisson Distribution

Normal Approximation to Poisson Distribution

If X is a Poisson random variable with E (X ) = λ and V (X ) = λ,
−λ
X√
Z = λ

is approximately a standard normal random variable. To approximate a binomial

probability with a normal distribution, a continuity correction is applied as follows:
 
P(X ≤ x) = P(X ≤ x + 0.5)Z ≈ P Z ≤ x+0.5−λ √
λ

and
 
x−0.5−λ
P(X ≥ x) = P(X ≥ x − 0.5)Z ≈ P Z ≥ √
λ

The approximation is good for λ > 5.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 41 / 43
Normal Approximation to Poisson Distribution

Example 1
The number of students who enroll in a psychology course is a Poisson random variable
with mean µ = 100. The professor in charge of the course has decided that if the number
enrolling is 120 or more, he will teach the course in two separate sections, whereas if
fewer than 120 students enroll, he will teach all of the students together in a single
section. What is the probability that the professor will have to teach two sections?
∞
100i
The exact solution: e −100
P
i!
: difficult to compute!
i=120
However, by recalling that a Poisson random variable with mean 100 is the sum of 100
independent Poisson random variables, each with mean 1, we can make use of the central
limit theorem to obtain an approximate solution.
If X denotes the number of students who enroll in the course, we have

P(X ≥ 120) = P(X ≥ 119.5) (the continuity correction)

= P( X√−100
100
≥ 119.5−100
√
100
) ≈ 1 − Φ(1.95) = 0.0256

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 42 / 43
Normal Approximation to Poisson Distribution

Example 2
Assume that the number of asbestos particles in a squared meter of dust on a surface
follows a Poisson distribution with a mean of 1000. If a squared meter of dust is
analyzed, what is the probability that 950 or fewer particles are found?

Let the random variable X denote the number of asbestos particles in a squared meter of
dust on the surface.
950
e −1000 1000x
P
P(X ≤ 950) = x!
x=0

The probability can be approximated as

 
950.5−1000
P(X ≤ 950) = P(X ≤ 950.5)≈ P Z ≤ √
1000
= P(Z ≤ −1.57) = 0.0582.

(Phan Thi Khanh Van) Chapter 3: Some special distributions February 19, 2025 43 / 43