4.17 Poisson distribution
Studying this course? Keep your progress in one place.
Key facts
- Poisson counts independent events at a uniform average rate in a fixed interval.
- , with .
- Rescale to the interval in the question before you calculate.
- Independent Poisson variables add: .
- Binomial is a count out of fixed trials; Poisson is a count at a random rate; normal is a continuous measurement.
When Poisson fits
Use Poisson for independent events occurring at a constant average rate in a fixed interval.
The Poisson distribution models the number of events in a fixed interval of time or space when:
- events are independent of one another, and
- events occur at a uniform average rate, and can occur at any point in the interval.
We write , where is the mean number of events in the interval.
Suitable
- Calls to an office per hour
- Flaws per square metre of fabric
- Typing errors per page
Not suitable
- Events in clusters, such as several patients from one accident
- A rate that changes through the day
- Events that trigger each other
Which situation is best modelled by a Poisson distribution?
Probabilities, mean and variance
Use the Poisson pdf and cdf on the GDC; the mean and variance are both .
The mean and variance are equal, both .
Use the GDC: Poisson pdf for , Poisson cdf for .
For use ; so .
Example: gives and . Formal proofs of the mean and variance are not required.
- Probability
- Mean and variance
. Which expression gives ?
Changing the interval
Rescale first: multiply the rate by the length of the new interval.
The parameter is the mean for the interval in the question. If the rate is per hour then for minutes , and for three hours (so the variance is also ).
Always rescale first, then use the distribution for the new interval.
- λ = 2
- λ = 12
- New λrate × length of interval
- Example per hour for hours:
Worked example
Patients arrive at per hour. Find the probability that more than arrive in two hours.
- 1
Rescale: two hours gives , so .
- 2
.
- 3
Use the Poisson cdf on the GDC.
Calls arrive at per hour. What is for minutes?
Sums of Poisson variables
The sum of independent Poisson variables is Poisson, with the means added.
If and are independent, then .
For joint events use independence: .
Only the sum is Poisson: is not.
- Po(1)
- Po(2)
- Po(3)
- Independent Poisson variables
Worked example
Flaws occur at and per square metre in two independent fabrics. Find the probability that one square metre of each has no flaws in total.
- 1
Total flaws .
- 2
.
and are independent. What is the distribution of ?
Binomial, Poisson or normal
Ask whether you are counting out of fixed trials, counting random events, or measuring.
Binomial counts successes in a fixed number of trials with constant . Poisson counts events occurring at random in a continuous interval, with no upper limit. Normal models a continuous measurement, symmetric about the mean.
"Out of bottles, each cracked with probability " is binomial; " bubbles per bottle on average" is Poisson; the mass of a loaf is normal.
- 1
Fixed number n of trials, probability p?
use binomial B(n, p)
- 2
Random events in a continuous interval?
use Poisson Po(λ), no upper limit
- 3
A continuous measurement?
use normal N(μ, σ²)
| Binomial B(n, p) | Poisson Po(λ) | Normal N(μ, σ²) | |
|---|---|---|---|
| What it models | Count of successes | Count of random events | Continuous measurement |
| Key clue | Fixed n trials, probability p | Average rate per interval | Mass, time, height |
Binomial B(n, p)
- What it models:
- Count of successes
- Key clue:
- Fixed n trials, probability p
Poisson Po(λ)
- What it models:
- Count of random events
- Key clue:
- Average rate per interval
Normal N(μ, σ²)
- What it models:
- Continuous measurement
- Key clue:
- Mass, time, height
A shop averages 2.5 customers per minute, arriving at random. Which distribution models the number of customers in a minute?
Try an exam question
Emails arrive at random at an average rate of per hour. (a) State two conditions needed for a Poisson model. (b) Find the probability that exactly emails arrive in one hour. (c) Find the probability that more than arrive in two hours.
[4 marks]
- [1]Events are independent and occur at a uniform average rate.
- [1], so .
- [1]For two hours , so .
- [1].
That's the notes covered.
Carry on to the next subtopic.
Try a question
Assuming has a Poisson distribution, use your GDC to find .
1 mark, marked against the mark scheme
Assuming has a Poisson distribution, use your GDC to find .
1 mark, marked against the mark scheme