8/12/2019 Lecturpoission distributione
1/12
The Poisson Distribution(Not in Reifs book)
8/12/2019 Lecturpoission distributione
2/12
The Poisson Probability Distribution
Simeon Denis Poisson "Researches on the probability
of criminal civil verdicts" 1837
Looked at the form of the
binomial distribution
When the Number of
Tr ials is Large. He derived the cumulative
Poisson distribution as the
L imiting case of theBinomial When the
Chance of Success
Tends to Zero.
8/12/2019 Lecturpoission distributione
3/12
The Poisson Probability Distribution
Simeon Denis Poisson "Researches on the probability
of criminal civil verdicts" 1837
Looked at the form of the
binomial distribution
When the Number of
Tr ials is Large. He derived the cumulative
Poisson distribution as the
L imiting case of theBinomial When the
Chance of Success
Tends to Zero.Simeon Denis Fish!
8/12/2019 Lecturpoission distributione
4/12
Poisson Distribution:An approximation to binomial
distribution for the SPECIAL CASEwhen the average
number (mean ) of successes is very much smal ler thanthe possible numbern. i.e.
8/12/2019 Lecturpoission distributione
5/12
The Poisson Distribution Models Counts.If events happen at a constant rate over time, the Poisson
Distributiongives
The Probabil i ty of X Number
of Events Occur r ing in a time T. This distribution tells us the
Probabil i ty of All Possible Numbers ofCounts, from 0 to I nf ini ty.
If X= #of counts per second, then the Poisson probability
that X = k(a particular count) is:
Here, the average number of counts per second.
!)(
k
ekXp
k
8/12/2019 Lecturpoission distributione
6/12
Mean and Variance for thePoisson Distribution
Its easy to show that for this distribution,
The Meanis:
Also, its easy to show that The Varianceis:.l
L
So, TheStandard Deviationis:
2
For a Poisson Distr ibution, thevariance and mean are equal!
8/12/2019 Lecturpoission distributione
7/12
Terminology: A Poisson Process
The Poisson parameter can be given as the meannumber of events that occur in a defined time period OR,
equivalently, can be given as a rate, such as = 2events per month. must often be multiplied by a timetin a physical process
(called a Poisson Process)
!
)()( k
etkXP
tk
= t = t
More on thePoisson Distribution
8/12/2019 Lecturpoission distributione
8/12
Example1. If calls to your cell phone are a Poisson process with a
constant rate = 2 calls per hour, what is the probabilitythat, i f you forget to turn your phone offin a 1.5 hour class,your phone rings during that time?
Answer: If X = #calls in 1.5 hours, we want
P(X 1) = 1 P(X = 0)
P(X 1) = 1 .05 = 95% chance
2.How many phone calls do you expect to get during the class? = t = 2(1.5) = 3
Editorial comment:The students & the instructor in the
class wil l not be very happy with you! !
05.!0
)3(
!0
)5.1*2()0(
330)5.1(20
eee
XP
8/12/2019 Lecturpoission distributione
9/12
9
Conditions Requiredfor the
Poisson Distr ibution to hold:
1. The rate is a constant, independent of time.2.Two events never occurat exactly the same time.
3. Each event is independent. That is, the
occurrence of one event does not make the nextevent more or less likely to happen.
8/12/2019 Lecturpoission distributione
10/12
Example
= (5 defects/hour)*(0.25 hour)= 1.25
p(x) = (xe-)/(x!)x = given number of defects
P(x = 0) = (1.25)0e-1.25)/(0!)
= e-1.25 = 0.287
= 28.7%
A production line produces 600 parts per hourwith
an average of 5 defective parts an hour. If you test
every part that comes off the line in 15 minutes,what is the probability of finding no defective parts(and incorrectly concluding that your process is perfect)?
8/12/2019 Lecturpoission distributione
11/12
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
Probability
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0
m
bino mialpo isso n 1
N=10,p=0. 1
0
0.1
0.2
0.3
0.4
0.5
Probability
0 1 2 3 4 5m
poisson
binomial
1
N=3, p=1/3
Comparison of Binomial & Poisson Distributions
with Mean = 1
Clearly, there is not much difference between them!
For N Large & m Fixed:
Binomial Poisson
N N
8/12/2019 Lecturpoission distributione
12/12
Poisson Distribution: As (Average # Counts) gets large,
this also approaches a Gaussian
l
= 5 = 15
= 25 = 35