Feature Selection Using Artificial Bee Colony and Discernibility Matrix in Rough Set Theory—A Hybrid Approach

In terms of time and space, machine learning algorithms are more sophisticated. The dataset's size is a key factor in lowering its complexity. The technique performs better and uses less storage space when the dataset is smaller. To attain promising performance of machine learning algorithms, a feature selection (FS) process is employed and the dimensionality gets reduced considerably. In FS, nature-inspired swarm intelligent algorithms have salient contributions. In this research, a hybrid feature selection algorithm is proposed by combining the Artificial Bee Colony (ABC) algorithm with the notion of discernibility matrix in rough set theory. The algorithm initially ranks all the features of the dataset with the discernibility-based Discrimination Frequency Relevance Measure (DFRM) algorithm. Then, FS is performed with ABC algorithm by considering the top-ranked features. To evaluate the utility of the generated feature subset, a classifier is employed. Using an existing ABC FS algorithm, the performance of the suggested technique is compared. The suggested hybrid technique works better than the current ABC algorithm, according to an experimental evaluation using datasets.