Multi-range progressive feature screening method

Abstract

The invention provides a multi-range progressive feature screening method, comprising following steps. Bootstrap is used to perform a leave-one-out method, including leave-one-out segmentation, training-validation segmentation, feature screening, and feature ranking. The leave-one-out segmentation divides a data set into a test set and a training set, and the training-validation segmentation performs a bootstrap segmentation in the training set with the leave-one-out. After that, a final feature selection is performed, including intersection feature screening and final feature set.

Images