Binomial and Poisson

Distributions
 Introduction
In the previous chapters, we have discussed the probability of particular
events. In many problems the ordinary laws of probability do not help to
compute the probability of an event. For example if a coin is tossed 20 times,
the number of possible outcomes will be (2)20 = 1048576 and suppose we
have to find the probability of getting heads. Practically it its difficult to write
down the sample space. Hence ordinary law(s) of probability do not help to
answer this question. Thus a special type of method is needed to find the
required probability. In such cases binomial distribution helps to find the
probability of required event. In addition to this there are occasions where the
number of trails is very large and the probability of occurrence of an event is
very small. In such cases binomial distribution takes a new form known as
Poisson distribution. It may be remembered that both binomial and Poisson
distributions are discrete type of distributions.
 Binomial Distribution
When a variable that has outcomes or values that tend to vary from
observation to observation because of chance related factors, it is called a
random variable. Some random variables can take on many values such as:
i) An individual’s test score can be any of the values say from 0 to 100.
ii) An individual can have any of the four martial status: single, married,
widowed and divorced. An individual can be 1 to 4 states.
There are variables which may have one of the two possible values or
outcomes.
i) Manufactured product is either defective or non defective.
ii) Answers in a test may be true or false.
iii) Aiming at target that comes in success or failure.
iv) Tossing a coin comes in H or T.
 Binomial Distribution
If variable has only two possible outcomes and if the probability of those
outcomes does not change for each trail then the variable is called a
“Binomial Variable”. For example the result of tossing a coin is a binomial
variable since probability of head and trail remains same for each toss.
The binomial distribution possesses the following properties:
i) The experiment consists of n repeated trails.
ii) Each trial results in an outcome that may be classified as success or
failure.
iii) The probability of success, denoted by p remains constant from trail to
trial.
iv) The repeated trails are independent.
 Binomial Distribution
To compute the binomial distribution, it is necessary to specify n; number of
trails x; the number of success, and p; the probability of success on each
trails.
The probability for binomial distribution is given by:
n
𝑃 𝑋 = 𝑥 = Cx 𝑝 𝑥 𝑞𝑛−𝑥 𝑓𝑜𝑟 𝑥 = 1,2,3, … , 𝑛
Where, 𝑞 = 1 − 𝑝, be the probability of an event that fails to occur.

Note: This distribution was developed by Bernoulli. Since this distribution

consists of binomial co-efficient nC x . Hence it is known as binomial
distribution.
 Bernoulli Trails: Assignment of Probabilities
Suppose that 𝑝 = 0.6, then 𝑞 = 1 − 0.6 = 0.4. We want to determine the
probability of getting 3 success in each 4 trails. There are 4 ways of getting
exactly 3 successes out of 4 trails.

Arrangements Probability
SSSF (0.6)(0.6)(0.6)(0.4)=0.0864
SSFS (0.6)(0.6)(0.4)(0.6)=0.0864
SFSS (0.6)(0.4)(0.6)(0.6)=0.0864
FSSS (0.4)(0.6)(0.6)(0.6)=0.0864
Total 0.3456
 Bernoulli Trails: Assignment of Probabilities
This is to be generalized by taking two things into consideration.
i) The number of ways the outcome can occur.
ii) The probability of one of these ways.
The number of ways of outcomes of which x are success and (𝑛 − 𝑥) are failure in
n trials is the permutation
n
of x objects of one kind (𝑛 − 𝑥) objects of another kind
and is given by Cx . The probability of one of these ways is
0.6 0.6 0.6 0.4 or (0.6)3(0.4)1
Thus, 𝑃 𝑋 = 3 = 4(0.6)3(0.4)1= 0.3456
Using the binomial distribution, we obtain
P ( X = x ) = nC x p x q n − x
With 𝑛 = 4, 𝑥 = 3, 𝑝 = 0.6 𝑎𝑛𝑑 𝑞 = 4
0.4,
𝑃 𝑋 = 3 = C3 (0.6)3(0.4)1= 0.3456
This is the same result as given above.
 Theorem
Prove that Binomial distribution is a probability distribution.
Proof: we have to show that
෍ 𝑃 𝑋=𝑥 =1
n
𝑤ℎ𝑒𝑟𝑒 𝑥 = 0,1,2, … . . 𝑛
𝑓𝑜𝑟 𝑎𝑙𝑙𝑥 =  nC x p x q n − x
x =0
n
=  nCx p x (1 − p )
n− x
where q = 1 − p
x =0

= C0 p (1 − p ) + C1 p (1 − p ) + C2 p (1 − p ) + ... + Cn p (1 − p )
n 0 n n 1 n −1 n 2 n−2 n n n−n

= (1 − p ) + C1 p (1 − p ) + ... + p n
n n n −1

= (1 − p + p )
n
by binomial exp ansion.
=1
 The Mean and Variance of a Binomial Random
Variable
We know that the mean of a random variable is defined as
𝜇 = n𝐸 𝑋
 =  x nC x p x q n − x
x =0
𝑛−1
+ 2 C2 𝑝2 𝑞𝑛−2 + ⋯ + 𝑛𝑝𝑛
n
= C1 𝑝𝑞
n

= 𝑛𝑝 𝑞𝑛−1 + 𝑛 − 1 𝑝𝑞𝑛−2 + ⋯ + 𝑝𝑛−1

= 𝑛𝑝 𝑞 + 𝑝 𝑛−1
Hence, 𝜇 = 𝑛𝑝 𝑎𝑠 𝑞 + 𝑝 = 1
 The Mean and Variance of a Binomial Random
Variable
Similarly, the variance of random variable is given by
𝜎 2 = 𝐸 𝑋 2 − 𝜇n2
E  X 2  =  x 2 nCx p x q n − x 2
Where, x =0
Replace 𝑥 by 𝑥 + 𝑥(𝑥 − 1), we obtain
n
E  X 2  =   x + x( x − 1)  nC x p x q n − x
x =0
n n
=  x Cx p q n x n− x
+  x ( x − 1) nC x p x q n − x
x =0 x =0
n n
n! n!
=x p q +  x ( x − 1)
x n− x
p x q n− x
x =1 x !( n − x ) ! x=2 x !( n − x ) !

= np  x
n
( n − 1) !
p q + + n ( n − 1) p 
x −1 n − x 2
n
( n − 2 )!
p x−2q n− x
x =1 ( x − 1)!( n − x )! x = 2 ( x − 2 ) !( n − x ) !

= np ( q + p ) + n ( n − 1) p 2 ( q + p )
n −1 n−2

E  X 2  = np + n 2 p 2 − n 2 p 2
 The Mean and Variance of a Binomial Random
Variable
Hence, 𝜎 2 = 𝐸 𝑋 2 − 𝜇2
= 𝑛𝑝 + 𝑛2 𝑝2 − 𝑛𝑝2 − 𝑛2 𝑝2
= 𝑛𝑝 1 − 𝑝 𝑏𝑢𝑡 1 − 𝑝 = 𝑞
𝜎 2 = 𝑛𝑝𝑞 , 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 1 − 𝑝 = 𝑞
 Example 01
The life length X of a certain type of resistor has a probability density
function given by,

=𝟎 , 𝒆𝒍𝒔𝒆 𝒘𝒉𝒆𝒓𝒆

With measurement of hundreds of hours. If five such resistors are

operating independently in a system, find the probability that at least one
will fail before 500 hours have elapsed.
 Solution
Solution:

−5Τ −5Τ
=− 𝑒 3 −1 = 1−𝑒 3 = 1 − 0.1888 = 0.8112
Now,
𝑃 𝑋 ≥1 =1−𝑃 𝑋 =0

−25Τ
=1−𝑒 3 = 0.999
 Example 02
The probability of passing soil compaction test is 0.80. If 15 tests were
performed at random, find the probability that
i) Exactly 5 compaction tests are passed
ii) From 3 to 5 compaction tests are passed
Solution: Here 𝑝 = 0.8, 𝑞 = 0.2 and 𝑛 = 15, then by binomial distribution,
𝑃 𝑋 = 𝑥 = Cx 𝑝 𝑥 𝑞 𝑛−𝑥
n

15
i) 𝑃 𝑋 = 5 = C5 (0.80)5(0.2)10= 0.000098
ii) 𝑃 3 ≤ 𝑋 ≤ 5 = 𝑃 𝑋 = 3 + 𝑃 𝑋 = 4 + 𝑃(𝑋 = 5)
Now, 𝑃 𝑋 = 3 = 15C3 (0.80)2(0.2)12= 0.000006
𝑃 𝑋 = 4 = C4 (0.80)4(0.2)11= 0.000011
15

15
𝑃 𝑋 = 5 = C5 (0.80)5(0.2)10= 0.000098
Hence, 𝑃 3 < 𝑋 < 5 = 0.000115
 Example 03
Large consignments of computer components are inspected for defectives
by means of a sampling system. Ten components are examined and the
lot is to be rejected if two or more are found to be defective. If a
consignments contains exactly 5% defectives. What is the probability
that the consignments is
(i) Accepted ii) Rejected
Solution: 𝑛 = 10, 𝑝 = 0.05, 𝑞 = 1 − 0.05 = 0.95
i) Now the consignment is accepted if
𝑃 𝑋 < 2 = 𝑃 𝑋 = 0 + 𝑃(𝑋 = 1)
n
Where, 𝑃 𝑋 = 0 = Cx 𝑝 𝑥 𝑞𝑛−𝑥
= C0 0.05 0 0.95 10
10

= 0.598
 Solution
And 𝑃 𝑋 = 1 = nCx 𝑝 𝑥 𝑞𝑛−𝑥
= 10C1 0.05 1 0.95 9
= 0.315
Thus the probability that the consignment is accepted= 0.598 + 0.315 =
0.913.

ii) The consignment is rejected if

𝑃 𝑋 ≥2 =1− 𝑃 𝑋 =0 +𝑃 𝑋 =1
= 1 − 0.598 + 0.315 = 0.087
 Poisson Distribution
Experiments yielding numerical values of a random variable X, the number of
outcomes occurring during a given time interval or in a specified region, are
often called as Poisson experiments. The given time interval may be of any
length, such as a minute, a day, a week or a year.

Examples:
i) The number of accidents per year in a particular factory.
ii) The number of faults in a length of cable.
iii) The number of cars crossing a bridge per hour in country side.
 Poisson Distribution
The number X of outcomes occurring in a Poisson experiment is called a
Poisson random variable and its probability distribution is called Poisson
distribution. The Poisson probability distribution depends on 𝜆, the average
number of outcomes occurring in a given time interval and is given by
𝑥
−𝜆 𝜆
𝑃 𝑋=𝑥 = 𝑒 for 𝑥 = 0,1,2, . . .
𝑥!
Where 𝜆 is a positive real number and 𝑒 ≈ 2.718.
This distribution was first developed by French mathematician Simeon D.
Poisson (1781-1840).
Note: A Poisson random variable has a (countably) infinite number of
possible values that are all non negative integars.
 Theorem
Prove that Poisson distribution is a probability distribution.
Proof: we have to show that sum of the probabilities is equal to 1, that is,
෍ 𝑃 𝑋=𝑥 =1
𝑓𝑜𝑟 𝑎𝑙𝑙𝑥
Now,

= 𝑒 −𝜆(𝜆0Τ0! + 𝜆1Τ1! + ⋯ )
= 𝑒 −𝜆(1 + 𝜆 + 𝜆2Τ2! + ⋯ )
= 𝑒 −𝜆. 𝑒 𝜆 = 𝑒 −𝜆+𝜆 = 𝑒 0 = 1
Hence Poisson distribution is a probability distribution.
 The Mean and Variance of a Poisson Random
Variable
Suppose random variable 𝑥 has a Poisson distribution with parameter 𝜆. Then
the mean and variance of can be obtained from the formula.
𝜇 = 𝜆 𝑎𝑛𝑑 𝜎 2 = 𝜆
Proof:

= 𝑒 −𝜆 (0 + 𝜆 Τ1! + 2𝜆2 Τ2! + 3𝜆3 Τ3! + … )

= 𝑒 −𝜆 𝜆(1 + 𝜆 + 𝜆2 Τ2! + 𝜆3 Τ3! + ⋯ )
= 𝑒 −𝜆 𝜆(𝑒 𝜆 ) = 𝜆
 The Mean and Variance of a Poisson Random
Variable
2 − 𝜇2
And ∞
𝑉 𝑋 = 𝐸 𝑋
𝑥 2𝑒 −𝜆 𝜆𝑥
𝐸 𝑋2 =෍
𝑥!
𝑥=0
∞
𝑥
𝜆
𝐸 𝑋2 = ෍[𝑥 𝑥 − 1 + 𝑥]𝑒 −𝜆
𝑥!
𝑥=0
∞ ∞
𝜆 𝑥 𝜆𝑥
𝐸 𝑋2 = ෍ 𝑥 𝑥 − 1 𝑒 −𝜆 + ෍ 𝑥𝑒 −𝜆
𝑥! 𝑥!
𝑥=0
∞ 𝑥=0
𝜆𝑥−2
𝐸 𝑋2 = ෍ 𝜆2 𝑒 −𝜆 + 𝜆
(𝑥 − 2)!
𝑥=2 ∞ ∞
2 2 𝜆𝑥−2 −𝜆 2 𝜆𝑥−2 −𝜆
𝐸 𝑋 =𝜆 ෍ 𝑒 + 𝜆, 𝑤ℎ𝑒𝑟𝑒, 𝜆 ෍ 𝑒 +𝜆= 1
(𝑥 − 2)! (𝑥 − 2)!
𝑥=2 𝑥=2
Thus, 𝐸 𝑋2 = +𝜆 𝜆2
Hence, 𝑉 𝑋 = 𝜆2 + 𝜆 − 𝜆2 = 𝜆
 Example 04
In a thin copper wire, suppose that the number of flaws follows a Poisson
distribution with a mean of 2.3 flaws per millimeter,
i) Determine the probability of exactly two flaws in 1 millimeter of wire.
ii) Determine the probability of 10 flaws in 5 millimetres of wire.
iii) Determine the probability of at least one flaw in 2 millimeter of wire.
Solution: i) Let 𝑥 denote the number of flaws in one millimeter of wire then,

ii) Average number of flaws per millimeter of wire is 2.3. Then the average
number of flaws in 5 millimetres of wire is 2.3(5)=11.5
Thus,
 Solution
iii) Let 𝑥 denote the number of flaws in two millimeter of wire.
Hence,

𝜇 = 2.3 2 = 4.6
𝑃 𝑋 ≥ 1 = 1−𝑃 𝑋 =0
= 1 − 𝑒 −4.6 = 0.9899
Poisson Approximation to the Binomial Theorem
When n is large, the binomial probability formula can be impractical to use
because of computational difficulties. Therefore, methods have been developed
that permit us to approximate binomial probability using formulas that are easy
to work with. One of those methods employs the Poisson probability formula
and applies when n is large and p is small (as a rule of thumb, we will require
that 𝑛 ≥ 100 and 𝑛 𝑝 ≤ 10 ).
In such cases we use Poisson distribution with parameter 𝜆 = 𝑛𝑝 to
approximate the binomial distribution.
𝑥
−𝑛𝑝 (𝑛𝑝)
Hence, 𝑃 𝑋=𝑥 = 𝑒
𝑥!
 Example 05
Ten percent of tools produced in a certain manufacturing process turn out
to be defective. Find the probability that in a sample of 10 tools chosen at
random
i) Exactly two will be defective.
ii) More than one will be defective.
Solution: Probability of defective tools= 𝑝 = 0.1 and 𝑛 = 10 given
Using Poisson Probability formula

Where 𝜆 = 𝑛𝑝 = 10 0.1 = 1
i)
 Solution
ii) 𝑃 𝑋 > 1 = 1 − 𝑃 𝑋 = 0 + 𝑃 𝑋 = 1
Where,

Hence 𝑃 𝑋 > 1 = 0.264