Bootstrapping is based on the jackknife method, which was proposed by Quenouille in 1949, and then refined by Tukey in 1958. The jackknife method is used for testing hypotheses and estimating confidence intervals. It's obtained by calculating the estimate after leaving out each observation and then computing the average of these calculations. With a sample of size N, the jackknife estimate can be found by aggregating the estimates of every N-1 sized sub-sample. It's similar to bootstrap samples, but while the bootstrap method is sampling with replacement, the jackknife method samples the data without replacement.

Bootstrapping is a powerful, non-parametric resampling technique that's used to assess the uncertainty in the estimator. In ...

Get Ensemble Machine Learning Cookbook now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.