WordGPT

Poisson distribution

Poisson distribution: the number of events in a fixed interval when events happen independently at a steady average rate. With the default parameters (average rate λ = 4) its mean is 4 and its standard deviation 2. Use the simulator below to draw samples and see it take shape.

Simulate it

Change the parameters, then draw random samples. The bars are the samples; the dots are the exact probabilities. With more samples the bars settle onto it.

Samples to draw

00.050.10.150.20246810121416x
Mean
sample - · exact 4
Standard deviation
sample - · exact 2

What is the Poisson distribution?

The Poisson distribution counts how many times something happens in a fixed stretch of time, distance or area when the events occur independently at a constant average rate λ. Emails arriving in an hour, typos on a page, calls to a help desk in a minute and defects along a roll of cable are all classic examples.

It has only one parameter, and a handy fingerprint: its mean and its variance are both equal to λ. If your count data has a variance much larger than its mean, a Poisson is the wrong model (try the negative binomial).

Where it is used

  • Customers or calls arriving per minute or per hour.
  • Typing errors per page.
  • Defects per metre of cable or per square metre of fabric.
  • Goals per football match (approximately).
  • Radioactive decays per second.
  • Rare events among very many trials: accidents per year, mutations per genome section.

Shape and parameters

For a small λ the distribution is piled up near 0 and skewed to the right; as λ grows it becomes more symmetric and approaches a normal curve with mean λ and standard deviation √λ. The most likely value is the whole part of λ (and λ − 1 as well when λ is a whole number).

What it assumes

  • Events happen independently of each other.
  • The average rate is constant over the interval.
  • Two events cannot happen at exactly the same instant.

How it relates to other distributions

  • Exponential distribution: The time between Poisson events is exponential.
  • Binomial distribution: With n large and p small, Binomial(n, p) is close to Poisson(np).
  • Gamma distribution: The time until the k-th Poisson event is gamma.
  • Negative binomial distribution: A Poisson with a varying rate gives an overdispersed negative binomial.

Common mistakes

  • Using a Poisson for counts whose variance is clearly bigger than the mean (clustered events).
  • Forgetting to scale λ to the interval: 4 per hour is 2 per half hour.
  • Assuming a constant rate when it clearly changes through the day.

How to work it out

  1. 1

    The probability mass function (the formula that gives the shape):

    P(X=k)=e−λλkk!P(X=k)=\frac{e^{-\lambda}\lambda^{k}}{k!}
  2. 2

    Its mean (the long-run average) and variance (the spread):

    μ=λ,σ2=λ\mu = \lambda,\qquad \sigma^2 = \lambda
  3. 3

    Worked example with the default parameters (average rate λ = 4): With λ = 4, what is the probability of more than 6 events?

    answer=0.1107\text{answer} = 0.1107
LaTeX source of the formulas on this page
\[ P(X=k)=\frac{e^{-\lambda}\lambda^{k}}{k!} \]
\[ \mu = \lambda,\qquad \sigma^2 = \lambda \]
\[ \text{answer} = 0.1107 \]

At a glance

PropertyValue
TypeDiscrete (counts)
Parametersaverage rate λ
Possible values0, 1, 2, …
Mean4
Variance4
Standard deviation2
Example (average rate λ = 4)With λ = 4, what is the probability of more than 6 events? Answer: 0.1107.

How to compute it in Excel, Python and R

ToolFunction
Excel / Google Sheets=POISSON.DIST(k, λ, FALSE) (TRUE for P(X ≤ k))
Python (SciPy)scipy.stats.poisson.pmf(k, lam)
Rdpois(k, lambda) (ppois for P(X ≤ k))

Questions

What is the Poisson distribution used for?

The number of events in a fixed interval when events happen independently at a steady average rate. Typical uses: Customers or calls arriving per minute or per hour. Typing errors per page. Defects per metre of cable or per square metre of fabric.

What are the mean and variance of the Poisson distribution?

Mean: μ = λ. Variance: σ² = λ.

What does the Poisson distribution assume?

Events happen independently of each other. The average rate is constant over the interval. Two events cannot happen at exactly the same instant.

How do I calculate it in Excel?

=POISSON.DIST(k, λ, FALSE) (TRUE for P(X ≤ k))

How is the Poisson distribution related to other distributions?

Exponential distribution: The time between Poisson events is exponential. Binomial distribution: With n large and p small, Binomial(n, p) is close to Poisson(np). Gamma distribution: The time until the k-th Poisson event is gamma. Negative binomial distribution: A Poisson with a varying rate gives an overdispersed negative binomial.

Related distributions

Choosing a distribution

Tables

    Poisson distribution: what it is, when to use it, formulas and simulator | WordGPT