PROBABILITY AND STATISTICS
Poisson Distribution
Poisson Distribution
Definition:
A variable x is said to have Poisson distribution if it takes the values with the probabilities
given by ! !
respectively.
Poisson distribution is generally derived as the limiting form of the
binomial distribution when and so that np is
finite quantity, say m. We come across cases in which the probability
of success is very small that is the probability of successes is less
than 0.1.
In general, if and , it can be taken to be a case of
Poisson distribution. Such events are known as rare events.
PROBABILITY AND STATISTICS
Poisson Distribution
Proof of Poisson’s Exponential Limit
By the Binomial distribution, f x n C x p x q n x
⇒ 𝑓(𝑥) Since lim (1 − ) =𝑒
1 2 𝑥−1 →
1− 1 − … (1 − ) 𝑛𝑝
𝑛 𝑛 𝑛 Where x is a finite quantity.
= (𝑛𝑥) (1 − )
𝑥! 𝑛
Hence, the probabilities of 0,1,2,…x,… success
Taking limits as 𝑛 → ∞, 𝑝 → 0 so that np = m, are given by
1
𝑓 𝑥 = 𝑚 𝑒 𝑒 . !
, !
, … respectively.
𝑥!
PROBABILITY AND STATISTICS
Poisson Distribution
Physical Illustration of Poisson Distribution
The following are some of the physical situations
illustrating Poisson distribution:
The number of deaths from a disease such as
heart attack
The number of printing mistakes per page of one of
the early proofs of a book
The number of defective materials per packing
manufactured by a good concern
The number of articles of a certain merchandise
sold by a concern in time t
The number of telephone calls received at a
particular switch board per time
PROBABILITY AND STATISTICS
Poisson Distribution
Poisson Process
A Poisson process is a model for a series of discrete events where the average time
between two events is known but the exact timing of the events is random. A Poisson
process meets the following:
Events are independent of each other. The occurrence of
one event does not affect the probability if another event
will occur or not.
The average rate (event per time period) is
constant.
Two events cannot occur at the same
time.
PROBABILITY AND STATISTICS
Poisson Distribution
Poisson Process
We can deduce a Poisson distribution without the help of
binomial distribution. Suppose X(t) is the number of random
calls received at a telephone exchange in time t. So, we make
the following assumptions:
• In a small interval of time dt, the probability of calls being
received is and the probability of no calls being received is
.
• The probability of more than one call being received in interval
dt is negligible.
• The probability of any call being received in time dt is
independent of previous calls.
PROBABILITY AND STATISTICS
Poisson Distribution
Characteristic Function for Poisson Distribution
The key parameter in fitting a Poisson distribution is the
mean value, which is usually denoted by m. This is the
average number of occurrences in a specified period. For
example: Cars passing in a minute on a highway.
In general, if X is a Poisson distribution, then
characteristics function is given by
!
for x = 0,1,2,…
PROBABILITY AND STATISTICS
Poisson Distribution
Moment Generating Function of Poisson Distribution
Let X be a discrete random variable with a Poisson
distribution and with parameter for some ,
then the moment generating function of random
variable X is given by
( )
PROBABILITY AND STATISTICS
Poisson Distribution
Mean of Poisson Process
The probability of the happening of x rare events is given by
Then expected value of Poisson distribution is given by
PROBABILITY AND STATISTICS
Poisson Distribution
Variance of Poisson Distribution
To find the variance of random variable X:
𝑣𝑎𝑟 𝑋 = 𝐸 𝑋 − [𝐸 𝑋 ] = 𝐸 𝑋 𝑋 − 1 + 𝑋 − 𝐸 𝑋
𝑣𝑎𝑟 𝑋 = 𝐸 𝑋 𝑋 − 1 + 𝐸 𝑋 − 𝐸 𝑋
=𝐸 𝑋 𝑋−1 +𝜆−𝜆
since 𝐸 𝑋 𝑋 − 1 = 𝜆 , therefore,
𝐸 𝑋 = 𝜆 −𝜆−𝜆 =𝜆
Hence, the variance of Poisson distribution is the same as
the mean of Poisson distribution.
PROBABILITY AND STATISTICS
Poisson Distribution
Mode of Poisson Distribution
Thus, if m is not an
Thus, for mode to be integer, then the
at x, mode is the integral
The value of x which
number lying between
gives the greatest
and m – 1 and m. If m is
probability is the ! !
an integer, then there
mode of Poisson
are two modes, one
distribution. ! !
each at m-1 and m.
Giving x < m < x + 1. The probability in both
cases is equal.
PROBABILITY AND STATISTICS
Poisson Distribution
Limiting Theorem: Poisson Distribution
Theorem: Show that Poisson distribution can be regarded as the limiting form of the binomial
distribution.
Proof: Poisson distribution is obtained as the limiting form of the binomial distribution. when 𝑛 →
∞, where np = m so that 𝑝 → 0.
The first four moments of binomial distributions are:
𝜇 = 𝑛𝑝, 𝜇 = 𝑛𝑝𝑞, 𝜇 = 𝑛𝑝𝑞(𝑞 − 𝑝), 𝜇 = 3(𝑛𝑝𝑞) + 𝑛𝑝𝑞(1 − 6𝑝𝑞)
The moments of the Poisson distribution can be deduced as
𝜇 = lim 𝑛𝑝 = 𝑚.
𝜇 = lim 𝑛𝑝𝑞 = lim(𝑛𝑝 1 − 𝑝 ) = 𝑚 since 𝑝 → 0
𝜇 = lim 𝑛𝑝𝑞 𝑞 − 𝑝 = lim 𝑚𝑞 𝑞 − 𝑝 = 𝑚 , since 𝑞 → 1 as 𝑝 → 0
𝜇 = lim(3 𝑛𝑝𝑞 + 𝑚𝑞 1 − 6𝑝𝑞 = lim 3𝑚 + 𝑚𝑞 1 − 𝑝𝑞 = 3𝑚 + 𝑚.
PROBABILITY AND STATISTICS
Poisson Distribution
Additive Property of Poisson Variates
Sum of independent Poisson variates is also a Poisson variates. Suppose are
independent Poisson variates with parameters , then is also a Poisson variate
with parameter .
Proof:
( ) for i=1,2,…,n
⋯ {since are independent}
( ) ( ) ( ⋯ )( )
⋯
Which is the moment generating function of a Poisson Variate with parameter
. Hence, the uniqueness theorem of moment generating function.
is also a Poisson variate with parameter .
PROBABILITY AND STATISTICS
Poisson Distribution
Example of Poisson Distribution
• Solution: The probability of
getting all the six heads in a
throw of six coins .
Example: Six coins are
tossed 6400 times. Using • Hence the mean is given by
the Poisson distribution,
what is approximate
probability of getting six • The probability of getting all
heads x times. six heads x times according
to Poisson distribution is
( )
!
PROBABILITY AND STATISTICS
Poisson Distribution
Theorem by Cumulants
Theorem: Show that for Poisson distribution with
parameter m, all the cumulants are m.
• Proof: Since the moment generating function of
Poisson distribution is ( ) , we have
( )