Error Estimation for Pattern Recognition

Abstract

The main training objective of the learning object is to introduce some of the most popular estimators for classification error: resubstitution, holdout, K-fold cross validation, bootstrap and repeated variants for holdout and K-fold cross validation. They are described at a basic level and then compared in terms of bias and variability.

Identifiers

DOI

10.1002/9781119079507

Citations and references