Importance sampling

From Simple English Wikipedia, the free encyclopedia

Importance sampling typically refers to a Monte Carlo method for sampling from a (target) distribution that cannot be sampled from directly. The method works by sampling from a proposal distribution (P(x)), which is ideally similar to the target distribution (Q(x)), and weighting each sample by the ratio of the likelihoods at the sampled point: w(x) = Q(x)/P(x).