Efficacy Of Feature Selectors For Classifiers In Opinion Mining

Emphasis placed on feature selection depends upon the learning algorithms. Feature selection identifies and removes irrelevant and redundant information and usually involves combining search and attribute utility estimation. It also evaluates the specific learning structures leading to many permutations. Data engineering is thought to be central in the mining applications development. Feature selection is an important and regularly used techniques in pre- processing the data for data mining as it reduces features number, removes irrelevant and noisy data and expedites data mining algorithms thereby improving mining performance parameters like accuracy and clarity. The aim of applying attribute evaluators for feature selection is the reduction of computational complexity and selected feature subsets improved classification accuracy. This paper aims to analyse different attribute evaluators/feature selectors and their classification using the K-Nearest Neighbour classifier following which evaluator performance is compared.