Hybridization of Arithmetic Optimization with Great Deluge Algorithms for Feature Selection Problems in Medical Diagnoses

Abstract

In the field of medicine, there is a need to filter data to find information that is relevant for specific research problems. However, in the realm of scientific study, the process of selecting the appropriate data or features is a substantial and challenging problem. Therefore, in this paper, two wrapper feature selection (FS) methods based on novel metaheuristic algorithms named the arithmetic optimization algorithm (AOA) and the great deluge algorithm (GDA) were used to attempt to tackle the medical diagnostics challenge. Two methods, AOA and AOA-GD were tested on 23 medical benchmark datasets. According to all of the experimental data, the hybridization of the GDA with the AOA considerably increased the AOA’s search capability. The AOA-GD method was then compared with two previous wrapper FS approaches;namely, the coronavirus herd immunity optimizer with greedy crossover operator (CHIO-GC) and the binary moth flame optimization with Lévy flight (LBMFO_V3). When applied to the 23 medical benchmark datasets, the AOA-GD achieved an accuracy rate of 0.80, thereby surpassing both the CHIO-GC and LBMFO V3.