TOPICS
Search

Categorical Distribution


The categorical distribution, also called the generalized Bernoulli distribution or multinoulli distribution, is the discrete distribution of a single outcome from k possible categories. If category i has probability p_i, where p_i>=0 and sum_(i=1)^(k)p_i=1, then a categorically distributed random variable X satisfies, for 1<=i<=k,

 Pr(X=i)=p_i.

The categorical distribution is the multinomial distribution with one trial. Equivalently, it can be represented by a vector having exactly one component equal to 1 and all other components equal to 0. For k=2, it reduces to the Bernoulli distribution after the two categories are labeled 0 and 1.


See also

Bernoulli Distribution, Discrete Distribution, Multinomial Distribution

Explore with Wolfram|Alpha

References

Murphy, K. P. Machine Learning: A Probabilistic Perspective. Cambridge, MA: MIT Press, 2012.

Cite this as:

Weisstein, Eric W. "Categorical Distribution." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/CategoricalDistribution.html

Subject classifications