Handbook of Statistical Analysis and Data Mining Applications

Chapter 5

Feature Selection

OUTLINE

Preamble

Variables as Features

Types of Feature Selections

Features Ranking Methods

Subset Selection Methods

Postscript

Preamble

After your analytical data set is prepared for modeling, you must select those variables (or features) to use as predictors. This process of feature selection is a very important strategy to follow in preparing data for data mining. A major problem in data mining in large data sets with many potential predictor variables is the curse of dimensionality. This expression was coined by Richard Bellman (1961) to describe the problem that increases as more variables are added to a model. As additional variables are added to a model, it may be able to predict a number better in regression ...