Poisson Distribution

The Poisson distribution is a discrete probability

distribution that applies to occurrences of some
event over a specified interval.

The random variable r is the number of

occurrences of the event in an interval. The
interval can be time, distance, area, volume, or
some similar unit.
 Poisson Distribution
Another commonly used discrete distribution.

The experiment results in outcomes that can be

classified as successes or failures.
Usually associated with rare events.

The probability that a success will occur is

proportional to the size of the region.

The probability that a success will occur in an

extremely small region is virtually zero.
 Poisson Distribution Requirements

• The average number of successes (µ) that occurs in a specified

interval is known.

• The occurrences must be random over an interval.

• The probability of the event occurring in an interval of space

or time is small.

• The occurrences must be independent of each other.

Parameter: Mean ()
 Poisson Distribution

The probability of r events occurring in a time period of t for a

Poisson random variable with parameter is:
𝑟
−𝜇 𝜇
𝑃 ( 𝑥= 𝑟 )=𝑒 ×
𝑟!

r represents expected number of events per unit of time.

represents mean expected number of events over time period
t.
 POISSON DISTRIBUTION
Only mean is required, from a posteriori observation.

The unique feature of Poisson:

Mean µ = variance

If we take 2 unit samples from a population which has 50%

defective cell phone, the mean would be
i.e. n = 2, p = 0.5, mean (=np) = 1
If we take 100 unit samples from a population
which has 1% defective cell phone, the mean
would be;

i.e. n = 100, p = 0.01, mean (=np) = 1

The above example shows that the same mean is

obtained with increasing sample size (n) and
decreasing probability (p)
 Poisson Distribution
The unique feature of Poisson:
Mean = variance

P (x=r)

0.8
 = 0.5
0.6  = 1.0
= 3.0

0.4
 = 6.0
0.2

10 15 r ( no. rare event)

5
Due to this, the mean can be estimated from a large
population (or sample) where the actual value is unknown
and the probability of the occuring event is small (also with
unknown value). This is the basis of Poisson distribution:

p → 0
n →
n, p and q are unknown

If r follows Poisson distribution, the probability of having r

event:
𝑟
−𝜇 𝜇
𝑃 ( 𝑥= 𝑟 )=𝑒 ×
𝑟!
 = mean of a rare event
r = no. event
e = 2.7183
Formula:

r: 0 1 2 3 4 …….... r

[ ]
0 1 2 3 4 𝑟
−𝜇 𝜇 𝜇 𝜇 𝜇 𝜇 𝜇
𝑒 + + + + + …+ ∨¿
0! 1! 2! 3! 4 ! 𝑟!

Where: P (X = 0)
P (X = 1)
P (X = 2)
P (X = 3)
Difference from a Binomial Distribution
The Poisson distribution differs from the binomial
distribution in these fundamental ways:

The binomial distribution is affected by the sample

size n and the probability p, whereas the Poisson
distribution is affected only by the mean μ.

In a binomial distribution the possible values of the

random variable x are 0, 1, . . . n, but a Poisson
distribution has possible x values of 0, 1, . . . , with no
upper limit.
 Poisson as Approximation to Binomial
• The Poisson distribution is sometimes used to
approximate the binomial distribution when n is
large and p is small.

• One rule of thumb is to use such an approximation

when the following two conditions are satisfied:
n ≥ 100
np ≤ 10
 Example 1

In a survey, the rate of accident in a state is 2 deaths/day.

Determine the probability of having no deaths/day and 3
deaths/day due to the road accident in that particular state.

𝑟
−𝜇 𝜇
𝑃 ( 𝑥=𝑟 )=𝑒 ×
𝑟!
( ) −2
𝑃 0 =𝑒 × 1=0.1353 dpois (0, 2)
3
−2 2
𝑃 ( 3 ) =𝑒 × =0.1804 dpois (3, 2)
3!
*In R, you can calculate probability for a Poisson distribution using dpois
(specify what is r and the mean)
Example 2

In a festive season, the rate of accident increases to 7

deaths/ day. What is the probability of having zero and
three deaths/day?

−7
𝑃 ( 0 ) =𝑒 ×1=0.0009
3
−7 7
𝑃 ( 3 ) =𝑒 × =0.0521
3!
Example 3
Mean no. of accident = 3 in 10 days
What is the probability of getting:
a) 5 accidents in 10 days (continously)?
5
3
−3
𝑃 ( 5 )=𝑒 × =0.101
5!

b) At least 1 accident in 10 days?

𝑃 ( 𝑟 ≥1 ) =1 − 𝑃 (𝑟 <1 )=1 − 𝑃 (𝑟 =0)

0
−3 3
𝑃 ( 𝑟 ≥1 ) =1 −𝑒 =0.95
0!
 Example 4
If the mean number of serious accidents per year in a large
factory is five, find the probability that in the current year there
will be:
a) No accidents
0
5
−5
𝑃 ( 0 ) =𝑒 × =0.0067
0!
b) Exactly three accidents

3
−5 5
𝑃 ( 3 ) =𝑒 × =0.1404
3!
c) At least three accidents
𝑃 ( 𝑟 ≥3 )= 𝑃 ( 3 )+ 𝑃 ( 4 )+ … 𝑜𝑟
𝑃 ( 𝑟 ≥3 )=1 −[ 𝑃 ( 0 ) + 𝑃 ( 1 )+ 𝑃 ( 2 ) ]

𝑃 ( 0 ) =0.0067
337

842

𝑃 ( 𝑟 ≥3 )=1 − [ 0.0067+ 0.0337+ 0.0842 ] =0.8754

 d) Four or more accidents

𝑃 ( 𝑟 ≥ 4 )=1 −[ 𝑃 ( 0 ) + 𝑃 ( 1 ) + 𝑃 ( 2 ) + 𝑃 ( 3)]

𝑃 ( 𝑟 ≥ 4 )=1 − [ 0.0067 +0.0337 +0.0842+ 0.1404 ] =0.7350

Example 5

In a certain population an average of 13 new cases of lung cancer

are diagnosed each year. Find the probability that in a given year
the number of newly diagnosed cases of lung cancer will be:

a) No more than 4
b) Fewer than 6
c) Between 9 and 11, inclusive
a) No more than 4

P(r 4) P(0)  P(1)  P (2)  P (3)  P(4)

 13 130
P (0) e  0.00000226
0!
 13 131
P (1) e  0.00002938
1!
132
 13
P (2) e  0.0001910
2!
 13 133
P (3) e  0.0008277
3!
 13 134
P ( 4) e  0.002690
4!
P(r 4) P(0)  P(1)  P(2)  P(3)  P(4) 0.0037
• b) Fewer than 6

P(r 5) P(0)  P(1)  P(2)  P(3)  P(4)  P(5)

P (0) 0.00000226

135
P (5) e  13  0.006994
5!
P(r 5) 0.00000226  ...  0.006994 0.0107
• c) Between 9 and 11, inclusive

P(9 r 11) P(9)  P(10)  P(11)

 13 139
P (9) e  0.06605
9!

 13 1310
P (10) e  0.08587
10!
11
13
P (11) e  13  0.1015
11!
P(9 r 11) P(9)  P(10)  P(11)

P(9 r 11) 0.06605  0.08587  0.1015 0.25342

 POISSON DISTRIBUTION OR NOT ??
To test whether a population follows Poisson distribution:
2
2 𝜎
𝜇=𝜎 =1
𝜇
Coefficient of Dispersion (C.D.)
The coefficient of variation measures variability in relation to the mean
(or average) and is used to compare the relative dispersion in one type
of data with the relative dispersion in another type of data.

2
If sample: 𝑠
=1
𝑥
 Application of Poisson
• The Poisson distribution has a number of uses in biology. One use is
to test whether the pattern of events in space or time is "random" or
not (spatial and temporal).

1) Seed example (spatial)

For example, imagine that you were to scatter seeds over a vast field
from a plane. Imagine also that you have divided the field up into blocks
of equal size, say 10 by 10 metres in area. If the probability that a given
square millimetre of soil receives a seed is low (you only dropped a few
thousand seeds), and if this probability is the same everywhere across
the entire field, and if seeds are independent of each other then the
number of seeds per block should follow a Poisson distribution.
• Here is what the Poisson distribution would look like if
the mean number of seeds per block is 2.0.
• If the probability is not the same across the entire field
or seeds not independent  non random patterns
1. Clumped – seeds are
aggregated (some
blocks have too many
seeds, some too few)
2. Overdispersed – seeds
distributed too evenly
across the field (each
block about the same
no)
 Rule of Thumb
• If CD ≅1 (σ2 = μ) Random distribution
• If CD > 1 (σ2> μ) Clustered distribution
• If CD < 1 (σ2 < μ) Uniform distribution
 Application of Poisson
2) Extinction events (temporal)
• Is the distribution of extinction events over
intervals of time across millions of years of fossil
history Poisson?
• Time interval of equal duration
 Application of Poisson
2) Extinction events (temporal)

Poisson?