Analysis of Feature Selection Algorithms on Classification: A Survey

Abstract

1. INTRODUCTIONmining is a process of knowledge discovery. The KDD is an automated process of knowledge discovery from the original data. The KDD consists of many steps like data cleaning, data integration, data selection, data transformation, data mining, pattern evaluation and knowledge representation. Among the steps the data selection is very much important to select the relevant feature and remove the irrelevant attributes. Classification is one of the datamining techniques used to discover the unknown class. The different classification methods in data mining are Bayesian classification (Statistical classifier), Decision tree induction, and Rule based classification (IF THEN Rule), Classification using Back propagation (Neural network algorithm), Support vector machine, Classification using Association Rule, k-nearest neighbor classifiers, casebased reasoning classifiers, Rough set approach, Genetic algorithm, Fuzzy set approach.