Optimized principal vectors-based co-kurtosis pca and recurrent attention unit-based Classifier for optimal cardiac disease prediction in big data

There is an enormous amount of data due to the latest advancements in contemporary data collection methods, instruments, and storage capacities. The number of features measured for each observation is indicated by the data's dimensions. High-dimensional data analysis has become a difficult task. The literature offers a variety of dimensionality reduction methods for getting rid of unnecessary characteristics. This paper develops an Optimized Principal Vectors-based co-Kurtosis Principal Component Analysis (Opv-KPCA) to obtain dimensionality reduction. In this phase, the necessary principal vectors are computed from high-order joint statistical moments obtained by using the co-kurtosis tensor. It is the best detect direction in the state space, which is defined as still dynamics. To select the optimal principal vector in the dimensionality reduction, use the Enhanced Egret Swarm Optimization Algorithm (EESOA). It is a combination of the Egret Swarm Optimization Algorithm (ESOA) and the Levy Fight (LF) algorithm. Finally, the Recurrent Attention Unit-based Classifier (RAUC) is developed for the classification of cardiac heart disease in patients. The performance of proposed approach is analyzed based on different metrics.