查看文章

Feature selection goal is to get rid of redundant and irrelevant features. The problem of feature subset selection is that of finding a subset of the original features of a dataset, such that an induction algorithm run on data containing only selected features makes a classifier to generate with the highest possible accuracy. High dimensional data can contain a high degree of irrelevant and redundant features which may greatly degrade the performance of learning algorithms. The performance of different feature selectors such as CFS, Chi-Square, Information Gain, Gain Ratio, One R and Symmetrical Uncertainty were evaluated on two different popular classification algorithms namely Decision Tree and Naive Bayesian method. A significant improvement in the performance of DT and NB classifier was shown after reducing the number of both irrelevant and redundant features by the use of different feature ranking methods.