The Poisson distribution
Studying this course? Keep your progress in one place.
Key facts
- counts random events at a constant average rate; is the mean for the interval.
- .
- Events must be independent, occur singly, and have a constant rate.
- Mean and variance both equal .
- In units the count is , and independent Poissons add: .
The Poisson model
It models the number of random events in a fixed interval, when they occur independently, singly and at a constant average rate.
A Poisson distribution models the number of times an event occurs in a fixed interval of time, length or area. Write , where is the mean number of events in the interval:
Events must occur independently, singly and at a constant rate. Examples are flaws in cloth, calls to a switchboard and misprints per page.
Which situation is least suitable for a Poisson model?
Calculating probabilities
Use the formula for one value and the tables for cumulative values, rewritten as .
For a single value use the formula or the calculator. The tables give , so rewrite other inequalities in that form. For whole numbers, "fewer than 4" means "at most 3".
This cumulative probability form stops you being one value out.
Worked example
. Find and .
- 1
.
- 2
From the tables, .
- 3
, because is the complement of .
. Which expression equals ?
Mean and variance
Both equal , so a sample with mean about equal to variance supports a Poisson model.
For the mean and the variance both equal the parameter. No derivation is needed. The standard deviation is .
A data set whose sample mean and variance are about equal supports a Poisson model. A variance much larger than the mean suggests events are clustered, so the model is poor.
- E(X)
- Var(X)
Worked example
A sample of counts has mean and variance . Is a Poisson model suitable?
- 1
For a Poisson model the mean and variance should be about equal.
- 2
Here the variance is about twice the mean.
. What is the standard deviation of ?
Scaling and adding
In units the count is , and independent Poisson counts add their means.
If events occur at rate per unit, then in units the number of events is . For example, 3 calls per minute becomes for 5 minutes.
The additive property: if and are independent, then . For non-overlapping intervals, treat counts as independent.
- Scaling in units
- Adding
Worked example
Flaws occur at a mean rate of per square metre of cloth. Find the probability of exactly flaws in square metres.
- 1
Rescale: in 5 square metres the mean is , so .
- 2
.
and are independent. What is the distribution of ?
Commenting on the model
Name the condition that fails and tie it to the situation.
Compare each condition with the context. A good answer names the condition and says how the context breaks it, for example "buses follow a timetable so arrivals are not random".
State the assumption when you use the model: "assuming flaws occur at random at a constant rate".
- 1
Name the condition
e.g. events occur at random
- 2
Compare with the context
does the situation really behave that way?
- 3
Say how it fails
e.g. buses follow a timetable
- 4
Conclude
the model is or is not suitable
Independence
- Could one event cause another, such as a breakdown causing a queue?
Singly
- Can events arrive in batches, such as a coach party?
Constant rate
- Does the rate change, such as rush hour against midday?
Customers arrive at a shop in large groups. Which condition fails?
Try an exam question
Flaws occur randomly in a roll of cloth at a mean rate of per square metre. Find the probability that a square metre piece has (a) exactly flaws, (b) more than flaws. State one assumption needed for the model.
[5 marks]
- [1].
- [1].
- [1].
- [1].
- [1]Assume flaws occur independently, singly and at a constant rate.
That's the notes covered.
Carry on to the next subtopic.
Try a question
Find .
1 mark, marked against the mark scheme
Find .
1 mark, marked against the mark scheme