ARTIFICIAL INTELLIGENCE FRAMEWORK FOR PREDICTING AND MANAGING EMPLOYEE ATTRITION IN IT INDUSTRIES

Abstract

In today’s volatile job market, retaining top talent has become a significant challenge for HR departments. The problem of high attrition rates, especially in tech-based firms, causes substantial loss in productivity and resources. This research aims to develop an AI-driven predictive model for talent retention by analyzing employee behavioral and performance patterns. Data was collected from five multinational IT firms in Bangalore, India, encompassing 3,500 employee records over three years. The dataset includes attributes like engagement scores, performance reviews, exit interviews, and training history. These were measured using structured questionnaires and internal HRMS logs. The analytical tool used was Python-based Scikit-learn with Random Forest Classifier. The proposed technique utilizes classification algorithms to segment employees into potential attrition risk levels. The proposed method includes a multi-level ensemble prediction framework trained on historical HR data. Six hypotheses were formulated, including H1—A significant relationship between training frequency and retention; H2—Performance ratings influence attrition risk; H3—Employee engagement scores are predictive of retention; H4—Work-life balance metrics impact employee loyalty; H5—Leadership feedback positively correlates with retention; H6—Salary dissatisfaction is a significant predictor of exit. The findings highlight key predictors and accuracy of the model in real-time application. The study concludes that AI-based predictive modeling significantly enhances HR decision-making and reduces turnover through targeted intervention.