AN ENSEMBLE FRAMEWORK OF STACKED MULTI REGRESSORS AND GRADIENT BOOSTING ALGORITHMS WITH TUNED HYPER PARAMETERS – ENSTAR TO PREDICT EMPLOYEE PERFORMANCE SCORE USING HR ANALYTICS

Machine learning (ML) supervised algorithms are immensely used in predictive analytics handling both categorical and numerical features. HR analytics is one such domain where complete data of employee details – known as historical data, are maintained which can be analysed to study the pattern of employee’s performance, attrition, absenteeism, career succession etc. to predict future outcomes. Regression and Gradient Boosting (GB) algorithms have proved to be versatile taking continuous values as input and predict the output. The objective of this research paper is to use the available HR data set from Kaggle, pre-process them, analyse the features’ importance and apply various regression algorithms as well as on the proposed ensemble stacking of regressors – EnStaR, to predict the employee performance score. Evaluation metrics like Mean Absolute Error (MAE), Mean Squared Error (MSE) and Coefficient of Determination (R2) are calculated and compared with all the individual algorithms as well as EnStaR to assert efficiency of the proposed model.