Enhancing Data Science Performance through PSO and GA-based Feature Selection on High-Dimensional Datasets

This research paper explores the application of Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) for data dimensionality reduction through feature selection. The study focuses on a dataset containing information about the dimensions of electronic components’ footprints. The rapid growth and complexity of high-dimensional datasets present challenges in effectively utilizing the information they contain. Dimensionality reduction and feature selection play vital roles in uncovering valuable information and enhancing the efficiency of data science applications. The paper applies PSO and GA algorithms to the dataset, with specific parameters set for each optimization technique. The results demonstrate that both PSO and GA are effective in feature selection, significantly reducing data dimensionality and improving classification accuracy. PSO outperforms GA in terms of mean accuracy, variance of accuracy, and execution time, making it a more efficient and effective optimization method for this specific classification problem.