Poisson Distribution: Rare Events Probability Explained

The Poisson distribution gives the probability of a given number of events happening in a fixed interval of time or space, when those events are rare and occur randomly at a known average rate. Typical examples: the number of phone calls arriving at a call centre in one hour, the number of printing errors on one page, or the number of cars passing a checkpoint in 10 minutes.

It matters because these situations do not fit the binomial distribution: there is no fixed number of trials n. Events just keep happening over time, and the Poisson formula needs only one parameter, the average rate, to produce every probability. This makes it the go-to model for queues, arrivals and defects per unit.

Key Concepts

• λ (lambda): the average rate of events per interval, the only parameter. It must be a positive number.
• k: the number of events, 0, 1, 2, 3 and so on, with no upper limit.
• e: Euler's number, approximately 2.71828, which appears in the formula.
• Mean and variance: both equal λ. So the standard deviation is √λ. This equality is a signature of the Poisson distribution.
• Assumptions: events occur one at a time, independently, at a constant average rate, and the interval is fixed.

Formulas

The probability mass function is P(X = k) = (e^(-λ) × λ^k) / k! for k = 0, 1, 2, …. The mean is μ = λ and the variance is σ² = λ. Two handy shortcuts: P(X = 0) = e^(-λ), and P(X = 1) = λe^(-λ). The Poisson is also the limit of the binomial when n is large and p is small with np = λ, which is why it suits rare events. Limits: if events cluster together (not independent) or the rate changes over the interval, the model will mislead.

Worked Example

A help desk receives an average of 4 calls per hour (λ = 4). What is the probability of exactly 3 calls in the next hour? Step 1: e^(-4) ≈ 0.018316. Step 2: λ³ = 4³ = 64. Step 3: k! = 3! = 6. Step 4: P(X = 3) = (0.018316 × 64) / 6 = 1.1722 / 6 ≈ 0.1954, or about 19.5%. For contrast, the probability of zero calls is e^(-4) ≈ 1.8%, and the mean and variance are both 4, so a typical hour has 4 ± 2 calls.

Common Misconceptions

• λ is not a probability: it is a rate, so it can be larger than 1 (for example 25 customers per hour). Probabilities stay between 0 and 1; λ has no such bound.
• The interval is fixed: if λ = 4 per hour, the rate for 30 minutes is λ = 2, not 4. Always rescale λ to the interval in the question before computing.
• Independence is required: if calls arrive in bursts because one event triggers others, the Poisson mean still matches but the variance is bigger than λ, and the formula underestimates extreme counts.

Real-World Applications

• Queues: banks and call centres staff shifts using Poisson arrival models for customers per hour.
• Manufacturing: modelling the number of defects per metre of fabric or per batch of chips.
• Insurance and finance: estimating how many claims or defaults arrive per month for pricing and reserves.

The Poisson distribution handles random arrivals with a single number, the rate λ: the formula e^(-λ)λ^k/k! gives every probability, the mean and variance are both λ, and scaling λ to the right interval is the most important step. Whenever events are counted over time or space rather than in fixed trials, Poisson is the natural model.