An Optimal Categorization of Feature Selection Methods for Knowledge Discovery

Abstract

With the continuous availability of massive experimental medical data has given impetus to a large effort in developing mathematical, statistical and computational intelligent techniques to infer models from medical databases. Feature selection has been an active research area in pattern recognition, statistics, and data mining communities. However, there have been relatively few studies on preprocessing data used as input for data mining systems in medical data. In this chapter, the authors focus on several feature selection methods as to their effectiveness in preprocessing input medical data. They evaluate several feature selection algorithms such as Mutual Information Feature Selection (MIFS), Fast Correlation-Based Filter (FCBF) and Stepwise Discriminant Analysis (STEPDISC) with machine learning algorithm naive Bayesian and Linear Discriminant analysis techniques. The experimental analysis of feature selection technique in medical databases has enable the authors to find small number of informative features leading to potential improvement in medical diagnosis by reducing the size of data set, eliminating irrelevant features, and decreasing the processing time.