HYBRID MACHINE LEARNING FRAMEWORK FOR PREDICTING EMPLOYEE ATTRITION

Abstract

Employee turnover worries companies since it affects operational stability as well as recruiting costs. Good attrition prediction will enable companies to develop proactive retention strategies that will eventually boost employee happiness and cut attrition. To forecast employee churn, this study introduces a Hybrid Machine Learning Framework that integrates classification, feature transformation, and data preprocessing into one design. Two different changes are suggested: SelectKBest with Mutual Information for feature selection and Truncated Singular Value Decomposition (TruncatedSVD) for dimensionality reduction. Three classifiers—Logistic Regression, Support Vector Machines (SVM), and K-Nearest Neighbors (KNN)—are used to assess these transformations in hybrid models. The SelectKBest_Logistic Regression hybrid performed the best, with an accuracy of 0.94 and a ROC-AUC of 0.97, which shows that hybrids based on feature selection do better than those based on dimensionality reduction. Key attrition predictors like job satisfaction, work-life balance, and distance from home are also found through feature importance analysis using SHAP values and permutation importance. By showing the effectiveness of hybrid models in predictive HR analytics, this research offers insightful advice for companies trying to maximize their employee retention policies.