ABSTRACT
In the big data feature selection research field, there have been various attempts are made to develop an efficient models for the selection of features in big data applications, due to the complicated nature of processing such data remains a major challenge. In this chapter, an exhaustive wrapper-based feature selection methodology is proposed. The proposed wrapper method of feature selection is parallelized on a multi-core CPU environment. The Euclidean separation matrices are used with the k-nearest neighbor (KNN) classifier for optimal solutions. The proposed work is focused on developing a more cost effective and reliable big data feature selection process. The proposed work aims to develop the method of feature selection such that it reaches the same level of classification accuracy and computation time as the traditional approach. An effective contribution aim is to assess the accuracy of six dataset results and to increase the accuracy value with fewer characteristics.
