Overview
Resampling is a technique for performing statistical inference of the measures of a sample of a dataset when the underlying distribution is unknown, or it is impractical to compute.
Resampling generates a number of new datasets from the original sample by randomly sampling from the dataset. Next, the statistic of interest is computed from the new datasets. A measure of the dispersion of the measured statistic, such as standard deviation, is computed from the new measures. This dispersion measure is then used to perform inference of the statistic.