Randomly sampling from a Distribution of data (typically with replacement, meaning that drawing samples does not affect the probability of drawing a similar sample in the future).