Map Reduce Framework-Assisted Feature Analysis and Adaptive Multiplicative Bi-RNN Using Big Data Analytics for Decision-Making

Abstract

In recent days, the usage of big data in different applications has improved rapidly, and also, it faces more complications due to enormous data. Generally, big data offers decision-making support to the decision-makers with high accuracy. The growth of communication and data contents is improved effectively according to the speed, velocity, size, and values for providing better knowledge to tackle upcoming complicated tasks and problems. On the other side, multi-criteria-aided decision-making technique is considered to tackle multiple problems presented in big data analysis. To achieve optimal outcomes, an automated model of big data analytics for improving the decision-making is proposed by utilizing the advanced methods. Initially, the big data is gathered from benchmark available sources. Consequently, the essential features are extracted based on the Map Reduce approach, where the features are analyzed by Spatial Incremental Principal Component Analysis (SI-PCA). Especially, in big data analytics, the Bidirectional Recurrent Neural Network (BiRNN) model facilitates increasing the overfitting issues that affects data quality. This issue is rectified by implementing the Adaptive Multiplicative BiRNN (AM-BiRNN) to enable accurate predictions to strengthen the decision-making performance. In the end, the resultant features are given as input to the AM-BiRNN. For further enhancement, the hyperparameters are optimally tuned by Improved Random Function-based Sculptor Optimization Algorithm (IRF-SOA). Finally, the validation of the model is done to achieve the high effective results. When compared with other state-of-the-art techniques, the impressive outcomes proved that the recommended system can provide a better decision-making outcome. Here, the experimental findings of the developed model show 93.15% of accuracy, and 87.09% of sensitivity, respectively.