Particle Swarm Optimization and Ensemble Methods for Absenteeism Prediction

Abstract

According to the High absenteeism among employees, this absenteeism can be detrimental to an organization as it can result in productivity and economic loss. machine learning (ML) techniques used to employee absenteeism of courier company. The study employs an inductive learning approach, leveraging classical data mining methods, including data preprocessing, feature selection using Particle Swarm Optimization (PSO), and classification through ensemble methods such as Bagging, Boosting, Voting, and Stacking. The dataset, consisting of 740 instances with 21 attributes, was preprocessed to remove irrelevant features and transformed for effective analysis. The proposed model achieved high accuracy, with Boosting emerging as the best performing method, achieving an accuracy of 99.8%. The study highlights the importance of feature selection in improving prediction accuracy and provides a valuable tool for managers to mitigate absenteeism's impact on organizational performance. Future work includes expanding the dataset, exploring advanced feature selection techniques, and integrating external factors for enhanced predictive capabilities