Geometric Distribution Negative Binomial Distribution Poisson Distribution

Statistics for Engineers (STAT3029A/CIVN2011A)

Lecture series 6b - Some special discrete distributions
Dr Honest Walter Chipoyera
School of Statistics and Actuarial Science
University of the Witwatersrand,
P Bag 3, Wits 2050
Republic of South Africa
E-mail: [email protected]
Date: September 11, 2023 1 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Outline

1 Geometric Distribution

2 Negative Binomial Distribution

3 Poisson Distribution

2 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Some Special Discrete Distributions

What this lecture series is all about

In this lecture series, you will be introduced to some of the most

common probability mass functions
Unit Objectives

At the end of this lecture series, you should be able to :

to state real-life settings where each of the following discrete
distributions may be applied: discrete uniform distribution,
Bernoulli and Binomial distributions, hyper-geometric
distribution, geometric distribution, negative binomial
distribution, Poisson distribution.
derive from first principles, the mean and variance of each of
the uniform, the Binomial and geometric distribution
3 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

The Geometric Distribution

The geometric distribution is applicable to a situation where

one makes independent trials until they obtain a success;
the probability of a success at each trial is θ; a failure occurs
with probability 1 − θ.
The number of trials needed for one to get a success is a
random variable X following a geometric distribution with
parameter θ.
Definition 1
A discrete random variable X is said to follow a geometric
distribution with parameter θ, i.e. X ∼ Geo(θ) if

 θ(1 − θ)x−1 , x = 1, 2, · · ·


fX (x) = P(X = x) = (1)

0, otherwise


4 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Theorem 1
Let X ∼ Geo(θ). Then

1 1−θ
E [X ] = and var (X ) = . (2)
θ θ2

To provePResult (2), one needs to start by evaluating

E [X ] = ∞x=1 x.θ.(1 − θ)
x−1 and E [X (X − 1)] using the following

results:

∞
X a
g (r ) = ar x = (3)
1−r
x=0
∞ ∞
X X a
g ′ (r ) = axr x−1 = axr x−1 = (4)
(1 − r )2
x=0 x=1
∞
X 2a
g ′′ (r ) = ax(x − 1)r x−2 = 5(5)
/ 13
(1 − r )3
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Example 1
A die is such that the odds of appearing of even numbered faces to
odd numbered faces is 2:1. If you decide to toss the die until you
get an odd-numbered face, what is the probability that you will
toss the die
a) 2 times? Getting an odd-numbered face can be viewed as a

x−1
success so that θ = 13 . P(X = x) = 31 × 23 and hence,
2
P(X = 2) = 9
b) at least 10 times?
∞  x−1
X 1 2 1 ( 2 )9 512
P(X ≥ 10) = × = × 3 2 =
3 3 3 1− 3 19683
x=10

6 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Negative Binomial Distribution

The Negative Binomial distribution applies to a random

experiment involving successive trials which are independent;
each trial results in a success (with probability θ) or a failure
(with probability 1 − θ).
The random variable of interest is the least number of trials
required to achieve k successes, X .
Remark The geometric distribution is a special case of the
Negative Binomial distribution in the sense that k = 1.
We use the notation X ∼ NB(k, θ) to mean that X follows a
Negative Binomial Distribution with parameters k and θ.

7 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Theorem 2
If X ∼ NB(k, θ), then
x−1
(1 − θ)x−k θk , x = k, k + 1, · · ·


 k−1
fX (x) = P(X = x) =
0, otherwise

(6)

Theorem 3
Let X ∼ NB(k, θ). Then

k k(1 − θ)
E [X ] = and var (X ) = . (7)
θ θ2

8 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Exercise
A large lot of tyres contains 10% defective. Four tyres are to be
selected for placement on a car.
a) Find the probability that six tyres must be selected from the
lot to get four good ones. Denote the number of trials until
you get 4 good tyres by X . Then X follows a negative
binomial distribution with k = 4 and θ = 0.9 and
 x−1
x−4 4
 4−1 0.1 0.9 , x = 4, 5, · · ·
P(X = x) =
0, otherwise


Thus P(X = 6) = 53 × 0.12 × 0.94 = 0.13122

b) Find the expected value and variance of the number of

selections that must be made to get four good tyres.
4
E [X ] = 0.9 and var (X ) = 4×0.1
0.92
9 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Poisson Distribution

The Poisson distribution typically models occurrences of a random

phenomena per unit of time or space. Examples are
the number of telephone calls a receptionist receives per hour
the number of mistakes in a randomly selected page of a book
the number of earthquakes that shakes Japan in the next
decade
etc.

10 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Definition 2
A random variable X is said to be Poisson distributed with
parameter λ (i.e. X ∼ Po(λ)) if and only if its probability mass
function is given by
 e −λ λx
 x! , x = 0, 1, 2, 3, · · ·
P(X = x) = (8)

0, otherwise

11 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Exercise

: Write down the Maclaurin series expansion of e r . Hence verify

that
∞
X e −λ λx
= 1.
x!
x=0

Theorem 4
Let X ∼ Po(λ). Then var (X ) = E [X ] = λ.

Remark 1
You are not required to be able to prove Theorem 4

12 / 13
Geometric Distribution Negative Binomial Distribution Poisson Distribution

Example 2
The number of requests for assistance received by a towing service
is a Poisson process with rate 3 per hour. The company has 4 tow
trucks and any calls received after all 4 trucks are booked are
referred elsewhere.
a) What is the probability that some calls are referred elsewhere
in the next 1 hour? For calls to be referred elsewhere it means
at least 5 requests for assistance would have been received.
 e −3 ×3x
P(X = x) = x! , x = 0, 1, 2, 3, · · ·
0, otherwise
P(X ≥ 5) = 1 − (p0 + p1 + p2 + p3 + p4 ) = 0.184737
b) Compute the probability that 5 requests are received during a
particular 2-hour period. Let the number of calls received in 2
hour period be denoted by Y . As will be discussed in Lecture
−6 5
series 7, Y ∼ Po(6) and P(Y = 5) = e 5!×6 = 0.1606231.
13 / 13