A hybrid feature selection method using multi-objective Jaya algorithm

Feature selection (FS) is a preprocessing procedure in machine learning that seeks to extract model predictors from a large geographic domain dataset to enhance prediction accuracy. They extract the most important features from a dataset and exclude those that could hinder estimation methods’ accuracy. However, as the number of observations grows, the dimensionality of the feature space grows, posing a considerable computational and prediction accuracy problem for numerous traditional feature selection approaches. In the context of software engineering, several feature selection strategies are studied. Normalized mutual information (NMI) and Jaya Algorithm are combined in this paper to create a hybrid feature selection algorithm. The proposed Normalized Mutual Information Jaya Algorithm (NMIJA) selection approach significantly decreases data dimension and minimizes classification redundancies. The reduced dataset achieves the highest classification accuracy when compared to typical feature selection algorithms. China, Desharnais, Kemer, and Maxwell are some of the datasets that were used. The numerical findings support the suggested model’s validity.