On the Efficacy of Boosting-Based Ensemble Learning Techniques for Predicting Employee Absenteeism

Abstract

AbstractEmployee absenteeism is a substantial problem faced by many organizations. It severely affects the productive operations in organizations. In the recent years, predictive modeling using ensemble learning techniques has increased the attention of researchers to develop competent models for various predictive tasks. Ensemble techniques combine the predictions of multiple models to yield a single consolidated decision. Thus, predictive modeling with the help of ensemble learning techniques can predict employee absenteeism so that the human resource department can devise intervention policies and lessen monetary losses due to absenteeism. Therefore, in this direction, this study develops models with the help of boosting-based ensemble learning aggregate with data balancing to predict employee absenteeism. The predictive accuracy of the absenteeism prediction models is evaluated using strong performance measures; area under receiver operator characteristics curve, balance and geometric mean. The results are also examined statistically with statistical analysis. The boosting-based ensemble learning techniques are effective for predicting employee absenteeism according to the results of the study.KeywordsMachine learningEnsemble learningData balancing