Skip to content
All apps

Poisson process

About this app

Rare, independent events with rate λ: arrivals on a timeline, a histogram of the event count against the Poisson PMF, exponential waiting times. The characteristic property mean = variance = λ — customer arrivals, claims, defects.

The statistics

A Poisson process describes rare, independent events that arrive at a constant rate λ per unit time, such as customer arrivals, calls, insurance claims or defects. The number of events in an interval of length 1 follows the Poisson distribution: P(k) = λᵏ e^−λ / k!. The waiting times between events are exponentially distributed with mean 1/λ. A defining property is mean = variance = λ. The Poisson distribution is the limit of the binomial distribution for n → ∞ and p → 0 with np = λ fixed.

What you see

At the top a realisation of the process unfolds as a sequence of events on a timeline. Below, a histogram shows how often the intervals contained 0, 1, 2 or more events, compared with the theoretical Poisson probability mass function. Alternatively the lower scene shows the distribution of waiting times with the exponential density λe^−λt. A values panel compares the mean and variance of the simulated counts with λ.

Things to try

Start with λ = 3 and notice that events are not evenly spaced: there are clusters and long gaps even though the rate is constant. Increase the number of intervals and the histogram approaches the Poisson distribution. Change λ: for small λ the distribution is skewed, for large λ it becomes more symmetric. Switch to waiting times and check that short gaps are the most common. Compare mean and variance in the values panel.

Frequently asked questions

What is a Poisson process?
A Poisson process is a model for randomly arriving events with a constant rate λ. Events occur independently, and no two happen at exactly the same moment. The count in a time interval is Poisson distributed, and the gaps between events are exponentially distributed.
Why are mean and variance equal in the Poisson distribution?
The Poisson distribution with parameter λ has mean λ and variance λ. This follows from its derivation as a limit of the binomial distribution: there the variance is np(1 − p), which tends to λ as p → 0 with np = λ fixed. If mean and variance differ clearly in real data, a Poisson model often does not fit.
What distribution do waiting times in a Poisson process follow?
Waiting times between consecutive events are exponentially distributed with density λe^−λt and mean 1/λ. The exponential distribution is memoryless: how long you have already waited does not change the distribution of the remaining wait. Short waiting times are the most common.
What can I set in the Poisson app?
You set the rate λ, the number of intervals and the animation speed, choose counts or waiting times for the lower scene and generate new realisations. The app compares the histogram with the theoretical distribution.

Subject: Statistics

More from Statistics

A VisuApp by heyprof: interactive, free in your browser, no sign-up. What is a VisuApp? · All apps