Geometric Distribution

BY DR. GARGI TYAGI

Lecture Highlights

• Concept and Definition

• Examples
• Mean and Variance
• Moment Generating Function
• Memoryless Property

BY DR. GARGI TYAGI

Binomial Distribution • X: Number of successes in finite number of trials

• X: Number of times an event occurs

Poisson Distribution
• Used for rare events

Negative Binomial
• X: Number of failures preceding 𝑟 𝑡ℎ success
Distribution

• X: Number of failures preceding first success

Geometric Distribution
• Negative Binomial Distribution with 𝑟 = 1

BY DR. GARGI TYAGI

Finding Probability of Number of failures preceding first success

The experiment is performed such that

1. Each trial has only two possible outcomes (say, success or failure).
2. The probabilities of success remain constant in each trial.
3. The trials are independent.

BY DR. GARGI TYAGI

 Geometric Distribution

A random variable 𝑋 is said to have a geometric distribution if it assumes only non-

negative values and its probability mass function is given by:
𝑃 𝑋 = 𝑥 = 𝑞 𝑥 𝑝, 𝑥 = 0,1,2, … , 𝑞 =1−𝑝
Where,
𝑋 = Number of failures preceding first success
𝑝 = Probability of success in each trial

Geometric distribution is a special case of Negative Binomial Distribution.

In other words, Negative Binomial distribution can be regarded as a generalization of geometric distribution.

BY DR. GARGI TYAGI

 Why the name Geometric Distribution?

Since the probabilities

𝑃 𝑋 = 𝑥 = 𝑞 𝑥 𝑝, 𝑥 = 0,1,2, … , 𝑞 =1−𝑝
are the terms of geometric progression.

BY DR. GARGI TYAGI

 Geometric Distribution
A candidate appears for a competitive exam until he qualifies the exam. Suppose the chances of his
qualifying the examination in any attempt is 0.3.
What is the probability that he would be successful at 4th attempt?
Example

BY DR. GARGI TYAGI

 Geometric Distribution
Mean and Variance

Mean
𝜇1′ = 𝐸 𝑋 = σ∞
𝑥=0 𝑥 𝑝 𝑥
𝑞
𝐸[𝑋] = σ∞ 𝑥 ∞
𝑥=0 𝑥 𝑞 𝑝 = 𝑝𝑞 σ𝑥=1 𝑥 𝑞
𝑥−1 = 𝑝𝑞 1 − 𝑞 −2 =
𝑝

Variance
𝑉[𝑋] = 𝐸 𝑋 2 − 𝐸 𝑋 2

𝐸 𝑋 2 = 𝐸 𝑋 𝑋 − 1 + 𝐸[𝑋]
2𝑞 2
𝐸 𝑋 𝑋−1 = σ∞
𝑥=0 𝑥(𝑥 − 1) 𝑞𝑥𝑝 = 𝑝 σ∞
𝑥=2 𝑥(𝑥 − 1) 𝑞𝑥 = 2𝑝𝑞2 1−𝑞 −3 =
𝑝

𝑞
𝑉𝑋 =
𝑝2

BY DR. GARGI TYAGI

 Geometric Distribution
A candidate appears for a competitive exam until he qualifies the exam. Suppose the chances of his
qualifying the examination in any attempt is 0.3.
What is the probability that he would be successful at 4th attempt?
What is the expected number of attempts he would require to qualify the exam?
Example

BY DR. GARGI TYAGI

Moment Generating Function

BY DR. GARGI TYAGI

 Geometric Distribution
Moment Generating Function

𝑀𝑋 𝑡 = 𝐸 𝑒 𝑡𝑋
𝑝
𝑀𝑋 𝑡 = σ∞
𝑥=0 𝑒 𝑡𝑥 𝑞 𝑥 𝑝 = 𝑝 σ∞ (𝑞𝑒 𝑡 ) 𝑥 =
𝑥=0 1−𝑞𝑒 𝑡

BY DR. GARGI TYAGI

Lack of Memory Property

BY DR. GARGI TYAGI

 Geometric Distribution
Lack of Memory Property

Suppose an event 𝐸 occur at one of the times 𝑡 = 0,1,2, … and the occurrence (waiting) time 𝑋 has
a geometric distribution with parameter 𝑝.
𝑃 𝑋 = 𝑡 = 𝑞𝑡 𝑝; 𝑡 = 0,1,2, …
Suppose we know that the event 𝐸 has not occurred before 𝑘, i.e. 𝑋 ≥ 𝑘.
Let 𝑌 = 𝑋 − 𝑘.
𝑌 is the amount of additional time needed for 𝐸 to occur.
What would be the probability that additional time required for the occurrence of event 𝐸 is 𝑡, i.e.
𝑃[𝑌 = 𝑡|𝑋 ≥ 𝑘]
Or, in other words, given that the event has not occurred till time 𝑘, what is the probability that it
would require additional time 𝑡 to occur.

BY DR. GARGI TYAGI

 Geometric Distribution
Lack of Memory Property

𝑋 ∼ 𝐺𝑒𝑜𝑚 𝑝
𝑃 𝑋 = 𝑡 = 𝑞𝑡 𝑝; 𝑡 = 0,1,2, …
Let 𝑌 = 𝑋 − 𝑘, is the amount of additional time needed for 𝐸 to occur.
Probability that additional time required for the occurrence of event 𝐸 is 𝑡, i.e.
𝑃[𝑌 = 𝑡|𝑋 ≥ 𝑘]

BY DR. GARGI TYAGI

 Geometric Distribution
Lack of Memory Property

𝑋 ∼ 𝐺𝑒𝑜𝑚 𝑝
𝑃 𝑋 = 𝑡 = 𝑞𝑡 𝑝; 𝑡 = 0,1,2, …
Let 𝑌 = 𝑋 − 𝑘, is the amount of additional time needed for 𝐸 to occur.
Probability that additional time required for the occurrence of event 𝐸 is 𝑡, i.e.
𝑃[𝑌 = 𝑡|𝑋 ≥ 𝑘]

BY DR. GARGI TYAGI

 Geometric Distribution
Lack of Memory Property

𝑿 ∼ 𝑮𝒆𝒐𝒎(𝒑) 𝑿 holds memory less property

BY DR. GARGI TYAGI

 Geometric Distribution
Lack of Memory Property

𝑿 ∼ 𝑮𝒆𝒐𝒎(𝒑) 𝑿 holds memory less property

BY DR. GARGI TYAGI

Relation between Geometric and Uniform Distribution

Let 𝑋1 and 𝑋2 be independent random variables each having geometric distribution

with parameter 𝑝, then the conditional distribution of 𝑋1 given 𝑋1 + 𝑋2 is uniform.

BY DR. GARGI TYAGI

Relation between Geometric and Uniform Distribution

Let 𝑋1 and 𝑋2 be independent random variables each having geometric distribution

with parameter 𝑝, then the conditional distribution of 𝑋1 given 𝑋1 + 𝑋2 is uniform.

BY DR. GARGI TYAGI

Summary

• Concept and Definition

• Examples
• Mean and Variance
• Moment Generating Function
• Memoryless Property
• Relation between Geometric and Uniform Distribution

BY DR. GARGI TYAGI

Thank You
BY DR. GARGI TYAGI