An Improved Machine Learning-Based Employees Attrition Prediction Framework with Emphasis on Feature Selection

Companies always seek ways to make their professional employees stay with them reduce extra recruiting and training costs. Predicting whether a particular employee may leave or not will help the company preventive decisions. Unlike physical systems, human resource problems cannot be described by scientific-analytical formula. Therefore, machine learning approaches are best tools for this aim. This paper presents three-stage (pre-processing, processing, post-processing) framework attrition prediction. An IBM HR dataset is chosen as case study. Since there several features in dataset, “max-out” feature selection method proposed dimension reduction pre-processing stage. implemented dataset. The coefficient of each logistic regression model shows importance results show improvement F1-score performance measure due method. Finally, validity parameters checked multiple bootstrap datasets. Then, average standard deviation analyzed check confidence value model’s stability. small indicates that stable more likely generalize well.