Sub-Feature Selection Based Classification

Feature selection by several mining techniques often issues with correlated features that generate the irrelevances for classification. Many existing approaches of feature selection may or mayn't remove the irrelevant data from data base for classification but few features don't need for feature selection schemes due to it involves same data in all classes. Based on this scenario, nothing is changed in traditional classes by different existing approaches. Thus the distinct classes are identified from traditional class using different sub-feature selection techniques. Hence it controls the redundancy and irrelevant features from individual database. The proposed model selects both features and sub-features for distinct classes using correlation coefficient and fuzzy model. The above model motivates to select the number of relevant features and sub-features towards distinct class without disturbing the original data set. It demonstrates the effectiveness of proposed model for selecting the number of features and sub-features from several data sets. The results reveal that the proposed method identifies the exact relevant features with sub-feature for novel class.