RoSaT– An Ensemble Architecture of Classification Algorithms to Predict Employee Rating Using SMOTEd IBM HR Analytics

Abstract

Performance evaluation of the employees reveal their productivity contributing to the growth of the organization they serve. It is therefore crucial for the HR department to ensure that there occurs no bias and the evaluation is done objectively. Machine learning techniques combined with appropriate tools can perform an exploratory data analysis and design a framework that predicts the performance of the employees without any human intervention. In this paper, we propose an ensemble stacking architecture – RoSaT that predicts the performance ratings with maximized accuracy. The dataset obtained was checked for data imputation followed by feature engineering process then training, validating and testing the data against different classification algorithms. The tabulated results along with performance metrics are finally elucidated to prove the efficacy of the proposed architecture.