Scala for Machine Learning - Second Edition

Bootstrapping with replacement

Bootstrapping is a Monte Carlo sampling technique to evaluate the sampling of a given distribution. This technique extracts samples from an independent and identically distributed dataset with a distribution that may not be known and represented as an empirical distribution function. The sample is created by selecting and replacing a data point, randomly chosen from the original population [8:5].

Overview

The purpose of bootstrapping is to estimate the accuracy of the resulting sampling by computing statistics characteristics such as mean, bias, standard deviation, average prediction errors, or confidence factors. As with any other sampling techniques, the resulting sample should be precise enough so that any statistical ...