Abstract
Big data is an important and complex dataset consisting of a large volume of data that helps to collect, store, and analyze data, depending on its applications and predictive analytics. During the predictive process, the method examines different quantities of data, which are difficult to process because their high dimensionality leads to difficulties in examining the correlations among the data. This paper introduces a method of optimized feature selection and soft computing techniques for reducing the dimensionality of the dataset. Initially, the data were collected from various resources that contained some inconsistent data, reducing the system's efficiency. Then, the inconsistent and noise data were removed by applying a normalized approach. Next, the optimized features were selected using the fireflies gravitational ant colony optimization (FGACO) approach. This optimized feature selection method successfully examines the characteristics and importance of the feature during the selection process. The selected feature consists of all details about particular predictive analytics. The system's efficiency was then evaluated using different datasets. The experimental results show that FGACO performs better in terms of the sensitivity, specificity, accuracy, and the number of selected features based on time.