Chapter 3: Some special distributions

Phan Thi Khanh Van

E-mail: [email protected]

April 9, 2021

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 1 / 48
Table of Contents

1 Bernoulli distribution

2 Binomial distribution

3 Hypergeometric distribution

4 Poisson distributions. Poisson processes

5 Uniform distributions.

6 Normal distributions. Central limit theorems

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 2 / 48
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) Chap 3: Some special distributions April 9, 2021 3 / 48
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) Chap 3: Some special distributions April 9, 2021 4 / 48
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) Chap 3: Some special distributions April 9, 2021 5 / 48
Binomial distribution

Flip 3 coins
Flip 3 coins, consider the number of heads:
x 0 1 2 3
P(X = x) 18 38 38 18

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) 0.008 0.096 0.384 0.512

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 6 / 48
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. Determine P(X = 2).
Outcome x Outcome x
OOOO 0 EOOO 1
OOOE 1 EOOE 2
OOEO 1 EOEO 2
OOEE 2 EOEE 3
OEOO 1 EEOO 2
OEOE 2 EEOE 3
OEEO 2 EEEO 3
OEEE 3 EEEE 4
P(X = 2) = 6 · P(E ) · P(O)2 = 6 · 0.12 · 0.92 = 0.0486.
2

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 7 / 48
Binomial distribution

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

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 8 / 48
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
PMF of X :
X 0 1 2 3 4
P(X = x) 0.6561 0.2916 0.0486 0.0036 0.0001
P
µ = Px.P(X = x) = 0.4.
σ 2 = x 2 .P(X = x) − µ2 = 4 × 0.1 × 0.9 = 0.36

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 9 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 10 / 48
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)
= C2521 .0.2521 0.754 +C 22 .0.2522 0.753 +C 23 .0.2523 0.752
25 25
+C2524 .0.2524 0.751 +C 25 .0.2525 = 9.6769.10−10 .
25
(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)
= C250 .0.7525 +C 1 .0.251 0.7524 +C 2 .0.252 0.7523 +C 3 .0.253 0.7522
25 25 25
+C254 .0.254 0.7521 = 0.2137.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 11 / 48
Hypergeometric distribution

Parts from Suppliers

A batch of parts contains 100 from a local supplier of tubing and 200 from
a supplier of tubing in the next state. If four parts are selected randomly
and without replacement
(a) what is the probability they are all from the local supplier?
(b) what is the probability that two or more parts in the sample are from
the local supplier?
(c) What is the probability that at least one part in the sample is from the
local supplier?

Let X equal the number of parts in the sample from the local supplier.
C 4 .C200
0
(a) P(X = 4) = 100 4
C300
= 0.0119.
(b) P(X ≥ 2) = P(X = 2) + P(X = 3) + P(X = 4) = 0.408.
(c) P(X ≥ 1) = 1 − P(X = 0) = 0.804.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 12 / 48
Hypergeometric distribution

Hypergeometric distribution
A set of N objects contains K objects classified as successes and N − K
objects classified as failures. A sample of size n objects is selected
randomly (without replacement) from the N objects where K ≤ N and
n ≤ N. The random variable X that equals the number of successes in the
sample is a hypergeometric random variable and
n−x
CKx .CN−K
f (x) = , x = max{0, n + K − N} to min{K , n}.
CNn

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 13 / 48
Mean and Variance
If X is a hypergeometric random variable with parameters N, K , and n,
then
µ = E (X ) = np and
σ 2 = V (X ) = np(1 − p) N−n
N−1 , where p = K /N.

Finite population correction factor

The term in the variance of a hypergeometric random variable
N−n
N−1

is called the finite population correction factor

If sampling were done with replacement, X would be a binomial

random variable and its variance would be np(1 − p).
If n  N, the correction is small and the hypergeometric distribution
is similar to the binomial distribution ⇒ a binomial distribution can
effectively approximate the hypergeometric distribution.
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 14 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 15 / 48
Example
A state runs a lottery in which 6 numbers are randomly selected from 40
without replacement. A player chooses 6 numbers before the state’s
sample is selected.
a) What is the probability that the 6 numbers chosen by a player match
all 6 numbers in the state’s sample?
b) What is the probability that 5 of the 6 numbers chosen by a player
appear in the state’s sample?
c) What is the probability that 4 of the 6 numbers chosen by a player
appear in the state’s sample?
d) If a player enters one lottery each week, what is the expected number
of weeks until a player matches all 6 numbers in the state’s sample?

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 16 / 48
Let X be the number of numbers chosen by a player that match the
numbers in the state’s sample. X is a hypergeometric random variable.
C66 C34
0
a) P(X = 6) = C406 = 2.6053.10−7
C65 C34
1
b) P(X = 5) = C406

C64 C34
2
c) P(X = 4) = C406

d) Assume that the ”success” is the event that the player matches all 6
numbers in the state’s sample.
Let Y be number of weeks until the first ”success”. Then Y is a
geometric random variable with the parameter
p = P(X = 6) = 2.6053.10−7 .
The expected number of weeks until a player matches all 6 numbers
in the state’s sample: E (Y ) = 1/p = 3838380.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 17 / 48
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) Chap 3: Some special distributions April 9, 2021 18 / 48
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) Chap 3: Some special distributions April 9, 2021 19 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 20 / 48
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) Chap 3: Some special distributions April 9, 2021 21 / 48
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 is a Poisson random variable
with parameter 0 < λ, and
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) Chap 3: Some special distributions April 9, 2021 22 / 48
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)
−10 100 101 102 103
=e 0! + 1! + 2! + 3! = 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) Chap 3: Some special distributions April 9, 2021 23 / 48
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)
−1 0 −1
1 − P(X = 0) − P(X = 1) = 1 − e 0!1 − e1! = 0.2642.
b) Let N be the number of cubic light-years must be studied and X be
N
the number of 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) Chap 3: Some special distributions April 9, 2021 24 / 48
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 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 = 13 , T2 = 32 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 )2
a) Therefore, P(X = 2) = 3
2! = 0.1982
b) 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 e 3 ( 20 )7
= P(X = 3).P(Y = 7) = 3
3! . 7!
3
= 0.0325.
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 25 / 48
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,
x−a
CDF: F (x) = 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 = 12

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 26 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 27 / 48
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

qthat an email message arrive.
√ (11−9) 2
q
1
a) σ= V = 12 = 3 (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 · 16 ) = 0.3333
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 28 / 48
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) = 0.5 · 3 · 41 = 0.375

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 29 / 48
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?

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 30 / 48
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 = 377
p 2 b−a
√
σ = V (x) = √ 12
= 3.
R 375
b) P(X < 375) = 374 dt 1
6 = 6.
R 380
c) P(X > x0 ) = 95% ⇔ x0 dt 6 = 0.95
380−x0
⇔ 6 = 0.95 ⇔ x0 = 380 − 6 · 0.95 = 374.3.
d) The mean extra cost:
C = (E (X ) − 375) · 0.002 = ( 374+380
2 − 375) · 0.002 = 0.004($)

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 31 / 48
Normal distributions
A random variable X with probability density function
−(x−µ)2
f (x) = √1 e 2σ 2 , −∞ < x < ∞
2πσ

is a normal random variable with parameters µ where −∞ < µ < ∞,

and σ > 0. Also,

E (X ) = µ and V (X ) = σ 2 .

and the notation N(µ, σ 2 ) is used to denote the distribution.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 32 / 48
For any normal random variable:
P(µ − σ < X < µ + σ) = 0.68
P(µ − 2σ < X < µ + 2σ) = 0.95
P(µ − 3σ < X < µ + 3σ) = 0.997

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 33 / 48
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: We often use table to evaluate the
13
integral.
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 34 / 48
Standard normal random variable
A normal random variable X with

µ = 0, σ2 = 1

is called a standard normal random variable and is denoted as Z .

The CDF of a standard normal random variable is denoted as

Φ(Z ) = P(Z ≤ z).

Standardizing a Normal Random Variable

If X is a normal random variable with
−(x−µ)2
E (X ) = µ, V (X ) = σ 2 , PDF: f (x) = √1 e 2σ 2 ,
2πσ
X −µ
the random variable Z = σ is a normal random variable with
E (Z ) = 0, V (Z ) = 1 : Z ∼ N(0, 1) : standard normal random
variable.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 35 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 36 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 37 / 48
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. What is the probability that a battery
lasts more than four hours?

Let X be the time until recharge for a battery in a laptop computer under
common conditions. X ∼ N(260, 502 )
Change the variable: Z = X σ−µ = X −260
50 , 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.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 38 / 48
Normal distributions

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?
d) If the standard deviation remains at 0.5 ounce, what must the mean
weight be 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.
−12
Change the variable: Z = X0.5 we have: Z ∼ N(0, 1).

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 39 / 48
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
⇒ σ = X Z−µ = 13−12
3.1 = 0.3226.
d) If σ = 0.5, then µ = X − Z σ = 13 − 3.1 · 0.5 = 11.45.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 40 / 48
Central limit theorem
Let X1 , X2 , · · · be a sequence of independent and identically distributed
(i.i.d.) random variables, each having mean µ and variance σ 2 .
Then the distribution of
X1 + ....Xn − nµ
√
σ n
tends to the standard normal N(0, 1) as 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) Chap 3: Some special distributions April 9, 2021 41 / 48
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. The normal distribution can be

used to provide an excellent approximation in this example.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 42 / 48
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) Chap 3: Some special distributions April 9, 2021 43 / 48
(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 44 / 48
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 and
 
X −160 150.5−160
P(X ≤ 150) = P(X ≤ 150.5) = P √ −5
≤√ −5
≈
160(1−10 ) 160(1−10 )
P (Z ≤ −0.75) = 0.227.

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

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 45 / 48
Normal Approximation to Poisson Distribution
If X is a Poisson random variable with E (X ) = λ and V (X ) = λ,
X√−λ
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) Chap 3: Some special distributions April 9, 2021 46 / 48
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
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) Chap 3: Some special distributions April 9, 2021 47 / 48
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
P e −1000 1000x
P(X ≤ 950) = x!
x=0

The probability can be approximated as

P(X ≤ 950) = P(X ≤  950.5)
950.5−1000
≈ P Z ≤ √1000 = P(Z ≤ −1.57) = 0.058.

(Phan Thi Khanh Van) Chap 3: Some special distributions April 9, 2021 48 / 48