Feature selection with Random Forest and Gradient Boosting
Álvaro Alonso Liso · 2016
The objective of the present work is to analyze the problem which arose naturally working with datasets with a large number of features, which usually forces the data analyst to select a small subset of all the available features to obtain acceptable training times and reduce over tting. The present work studies the usefulness of the feature importance coe cients given by Trees, Random Forest and Gradient Boosting regressors applied to a problem of wind energy production.