An Enriched Employee Retention Analysis System with a Combination Strategy of Feature Selection and Machine Learning Techniques

In today’s competitive business environment, retaining regular employees is a major challenge for all organizations. Retention of employees is an essential management strategy that hopes to create positive work environment and achieve lasting success. Thus, researchers turned their attention to machine learning framework to quickly find the reasons of lowering attrition and estimate staff turnover. In reality, many factors influence employees’ decisions to remain with the organization. The high dimensionality of employee data makes it difficult for data analysts to comprehend and analyses in any meaningful way within time. As a consequence, feature selection got a special focus in the research in order to improve the quality of the data analysis and discover the pertinent data. In this study, the authors propose an Enriched Employee Retention Analysis System (EERAS) that uses a novel approach with a right combination of feature selection model and machine learning techniques. The double goal of the proposed EERAS is to (1) identify the factors that influence employee turnover and (2) recommend key factors for employee retention. To achieve the aimed goals, the authors undertake an empirical analysis on a large employee dataset with four feature selection strategies: MRMR Chi-2, ANOVA and Kruskal on five different classification learning algorithms - Decision Tree, Linear Regression, Naive Bayes, Support Vector Machine, and XGBoost. The results demonstrated that feature selection improves classification accuracy while decreasing training time. In addition, the system recommends critical elements that might be incorporated into HR policy to ensure long-term staff retention.