Adaptive Feature Selection and Classification Using Optimization Technique

Abstract

Today, the dimensions of the datasets are becoming more and more. From the existing or collected features, some are useful and some are not useful. In order to analyze the dataset, a feature subset can be selected by applying feature selection method. To select best features in lesser time, optimization algorithms like Particle Swarm Optimization (PSO) can be implemented. There is a possibility of redundant features getting selected based on the correlation among features. Pearson correlation coefficient is being employed to reduce the redundancy. Further, PSO is implemented for classification. Datasets analyzed in this research work are, namely, Australian credit, German credit, Iris, Thyroid, Vehicle, WBC, and Wine. It is observed that convincing results were obtained from the proposed method.