AN APPROACH FOR UNSUPERVISED FEATURE SELECTION USING GENETIC ALGORITHM

Each data mining application has widespread issue; dataset has gigantic number of features which are irrelevant or redundant to the data mining task in hand which negatively affects the performance of the elementary learning algorithms, and makes them less er capable. There is difficulty of inadequate increase in dimension is strappingly related to fascination of cassette or measuring data at a far granular level then it was done previously. There is no doubt that this is a blistering problem. It has started gaining more magnitude recently due to surge in data. Hereafter plummeting the dimensionality of dataset is principal and imperative job for data mining applications and machine learning algorithms in order that computational burden of the learning algori thms can be minimized. In this paper we will measure up to the GFS (Greedy Feature Selection) and our proposed method and diverse unsupervised feature selection algorithms discussed in order to find out factors which influence the performance of existing a lgorithm. In our proposed method we have incorporated the Genetic feature selection method and GFS and TPR (True Positive Rate), FNR (False Negative Rate) estimated using KNN Classifier.