FEATURE SELECTION APPROACHES WITH TEXT MINING FOR CATEGORICAL VARIABLE SELECTION

Abstract

Feature Selection is the process of selecting a subset of relevant features for use in model construction. The Feature Selection methods are used to increase the overall efficiency of the classification model. The amount of text data is increasing rapidly in recent years, the feature selection approaches are important for the preprocessing textual documents for data mining. The feature selection method focuses on identifying relevant data that help to reduce the preprocessing of
huge amount of data and reduce the data size by removing irrelevant or redundant attributes. The feature selection algorithm conducts a search for best subset using valuation algorithm. The valuation algorithm is run on the dataset with different set of features removed from the data. The main objective of this paper is to improve the accuracy of classification of text documents by removing the irrelevant, noisy features and compare the precision and recall of various Feature Selection methods. The performance of Feature Selection methods with various  classifiers are compared and tabulated.