Generalized wait-times for multiple events
The gamma distribution extends the exponential distribution to model the time until k events occur in a Poisson process. When k = 1, the gamma distribution reduces to the exponential; when k > 1, it shifts probability rightward, requiring multiple events before the observation time is recorded.
The distribution is parameterized by a shape parameter k (number of events) and a rate parameter lambda (event rate). Larger k values produce distributions that pile up around their mean and have narrower tails; smaller k (closer to 1) produce sharp peaks near zero with fat right tails.
Versatility across domains
The gamma distribution appears in diverse fields. In reliability engineering, it models failure time of systems with components in series. In meteorology, it describes rainfall accumulation. In finance, it models insurance claim sizes and portfolio drawdowns.
The shape parameter k provides a biologically or physically meaningful interpretation: for rainfall, k might represent the number of cloud passages; for insurance, it captures the complexity of claim profiles. This interpretability makes gamma distributions popular for Bayesian priors in domains where experts can reason about event counts.