TOPICS
Search

Exponential Family


An exponential family is a collection of statistical distributions whose probability densities or mass functions can be written

 f(x|eta)=h(x)exp[eta^TT(x)-A(eta)].

Here eta is the natural parameter, T(x) is a vector of sufficient statistics, and A is the log-partition function that normalizes the distribution.

When differentiation under the integral is valid, the gradient and Hessian of A give the expectation value and covariance matrix of T(X). The normal distribution, Poisson distribution, binomial distribution, and gamma distribution have standard exponential-family forms, although the choice of parameterization and support matters.


See also

Binomial Distribution, Gamma Distribution, Normal Distribution, Sufficient Statistic

Explore with Wolfram|Alpha

References

Casella, G. and Berger, R. L. Statistical Inference, 2nd ed. Pacific Grove, CA: Duxbury, 2002.

Cite this as:

Weisstein, Eric W. "Exponential Family." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/ExponentialFamily.html

Subject classifications