A Hybrid Big Data Analytics Framework for Business Intelligence Classification Using Whale Optimization Algorithms and Neural Network Models

Abstract

Big data plays a significant role in the information industry, and the vast data generation is witnessed by digitization. The process of analysing and handling the issues of velocity, volume, variety, and veracity of the data with the assistance of traditional approaches are ineffective. To overcome the shortcomings of the traditional and other mining approaches, deep learning has initiated to handle the issues of big data. In this research article, the whale optimization algorithm (WOA) is incorporated with an artificial neural network (ANN) for classifying the imbalanced dataset. Thewhale optimization algorithm with the neural network approach (WONNA)is composed of whale optimization, selection of features, pre-processing, and classification. In the optimization phase, whale optimizationremoves the redundancy and irrelevancy of the features where the accuracy of the classification is enriched. The synthetic minority oversampling technique (SMOTE) and synthetic minority oversampling technique with rough set theory with a subset lower approximation (SMOTE-RS B*) is used in the pre-processing and the data is classified in the classification phase. The experimental result shows that the proposed scheme with SMOTE pre-processing has effective performance and promising results when compared to the existing approaches.