A Novel Cluster of Quarter Feature Selection Based on Symmetrical Uncertainty
Sai Prasad Potharaju, M.A. Sreedevi · DergiPark (Istanbul University) · 2018
Due to the diversity of sources, a large amount of data is being produced and also has variousproblems including mislabeled data, missing values, imbalanced class labels, noise and highdimensionality. In this research article, we proposed a novel framework to address highdimensionality issue with feature reduction to increase the classification performance of variouslazy learners, rule-based induction, bayes, and tree-based models. In this research, we proposedrobust Quarter Feature Selection (QFS) framework based on Symmetrical Uncertainty AttributeEvaluator. Our proposed technique analyzed with Six real world datasets. The proposedframework , divide whole data space into 4 sets (Quarters) of features without duplication. Eachsuch quarter has less than or equals 25 % features of whole data space. Practical results recordedthat, one of the quarter, sometimes more than one quarter recording improved accuracy than thealready available feature selection methods in the literature. In this research, we used filter-basedfeature selection methods such as GRAE, IG, CHI-SQUARE (CHI 2), Relief to compare thequarter of features produced by proposed technique.