A type of discrete distribution describing the ubiquity and rarity of values in a data set. E.g. the frequency of words in a language form a Zipf distribution, with a few words such as "the" and "you" occurring very frequently, a somewhat larger group of words which occur fairly regularly such as "chair" and "watch", but with the majority of words being very rare, such as "sonorous" or "latibule".

The frequency of the $k$-th most common value follows $f(k)\propto \frac{1}{k^s}$. The Zipf distribution assumes that the decrease is constant, e.g. each word is n% more rare than the next most common one.