Learn & Understand

The Poisson Distribution and the Law of Rare Events

The companion calculator computes Poisson probabilities for events that arrive at a steady average rate, like calls to a support line or typos on a page. The Poisson distribution has a rich history as the "law of rare events," and it arises from a beautiful connection to the binomial distribution. Understanding where the Poisson comes from as a limit of the binomial, the famous data set that first showcased it, and the conditions it assumes turns a Poisson calculation into an appreciation of the distribution that governs rare, randomly occurring events.

Modeling Events Without a Fixed Number of Trials

The Poisson distribution fills a gap the binomial cannot: it models the count of events that occur over an interval of time or space where there is no natural fixed number of trials, only an average rate. Some situations have countable trials (fixed inspections, fixed flips), but others do not: calls arriving over an hour, customers entering over a day, defects along a length of material, there is no obvious "number of trials," just events happening at some average rate over the interval, as the calculator's context notes. The Poisson distribution requires only that average rate to give the probability of any particular count in the interval, which is what makes it so useful for arrival-type processes. This is the Poisson's distinctive niche: it describes how many events occur in a continuous span given only their average rate, without needing a fixed number of trials. Understanding that the Poisson models rate-based counts rather than fixed-trial counts is the key to when it applies: whenever events occur randomly over an interval at a roughly steady average rate, the Poisson gives the distribution of how many will occur. The calculator's single input, the average rate, reflects this, since the rate is all the Poisson needs.

The Poisson as a Limit of the Binomial

The Poisson distribution has a deep origin: it arises as a limiting case of the binomial distribution when the number of trials is very large and the probability of success on each is very small.

Imagine dividing an interval into a great many tiny sub-intervals, in each of which an event either occurs (with small probability) or does not, this is like a binomial with a huge number of trials and a tiny success probability, but with the average number of events (rate times interval) held fixed. As the sub-intervals become infinitely fine, the binomial converges to the Poisson distribution, which is why the Poisson is often derived as this limit. This explains why the Poisson models rare events: it is the distribution that emerges when there are many opportunities for an event but each is individually unlikely, so events occur sparsely at some average rate. It also connects the two distributions the calculator offers: the binomial for fixed trials, the Poisson for its many-trials, rare-event limit, as the calculator's cross-reference notes. Understanding the Poisson as a limit of the binomial reveals its origin and its domain: it is the natural distribution for counting rare, randomly scattered events over an interval, arising from the binomial when opportunities are many and each success is rare. This is why it is called the law of rare events, it governs how many rare events occur when there are many chances for them.

The Famous Horse-Kick Data

The Poisson distribution's power was memorably demonstrated by one of the most famous data sets in the history of statistics: the number of soldiers killed by horse kicks in Prussian army corps over many years. This unusual data, deaths from horse kicks were rare, random events occurring at a low average rate, was found to follow the Poisson distribution remarkably well, providing a striking real-world confirmation that the law of rare events describes such phenomena. The horse-kick example became a classic illustration precisely because it is so vivid and unexpected: a grim, seemingly random occurrence follows a clean mathematical law, showing that even rare, irregular events obey the Poisson distribution when they occur independently at a steady average rate. This historical example established the Poisson as the model for rare events across countless fields, from accidents and failures to disease cases and radioactive decays, all of which are rare events occurring at some average rate. Understanding the horse-kick data illustrates what the Poisson captures: the pattern of rare, independent events over many opportunities, which recurs throughout nature and human affairs. The example endures because it shows the Poisson's surprising reach, that a distribution derived as a mathematical limit accurately describes real rare events, which is exactly the kind of phenomenon the calculator models, whether calls, failures, or defects.

The Conditions the Poisson Assumes

Like any model, the Poisson rests on assumptions, and understanding them clarifies when it applies. It assumes events occur independently, one event does not make another more or less likely, and at a constant average rate over the interval, the rate does not change within the span being modeled. It also assumes events occur one at a time (not in simultaneous clusters) and that the rate is proportional to the interval length. When these conditions hold, events happening independently, randomly, and at a steady average rate, the Poisson gives the correct distribution of counts, and its notable feature that the mean and variance are both equal to the rate holds. When they fail, if events cluster, if the rate varies over the interval, or if events are not independent, the real distribution may be more spread out or otherwise depart from the Poisson, so the model can mislead, much as the binomial fails when its assumptions break. Overdispersion, more variability than the Poisson predicts, is a common sign that events are clustering or the rate is varying. Understanding the conditions the Poisson assumes is essential to applying it responsibly: it is the right model for independent, steady-rate rare events, but not for clustered or rate-varying processes. The calculator computes Poisson probabilities assuming these conditions; understanding them is what reveals whether a real rate-based process genuinely follows the law of rare events or departs from it. This is the same discipline of checking assumptions that governs all probability models.

Understanding the Poisson Distribution

Use the calculator to compute Poisson probabilities, and understand the distribution behind them: the Poisson models the count of rare events over an interval given only their average rate, arising as the limit of the binomial when trials are many and success is rare, which is why it is the law of rare events, famously confirmed by the Prussian horse-kick data. It assumes independent events at a steady rate. The calculation gives the probability of a count; understanding the Poisson's origin and assumptions is what reveals when rare-event processes follow it.