A Web-Based Machine Learning Application for Employee Attrition Prediction

Employee attrition is a critical issue faced by organizations, a major challenge faced by organizations worldwide, directly impacting workforce stability, productivity, and operational costs. Traditional methods of analyzing attrition often rely on descriptive statistics, which may not fully capture complex interactions among demographic, behavioral, and organizational factors. This paper proposes a machine learning approach to predict employee attrition using a Random Forest Classifier with both numerical and categorical features, presents a balanced approach combining machine learning and web-based deployment for predicting employee attrition using the IBM HR Analytics dataset. The model achieved an accuracy of 87.41%, indicating its effectiveness in identifying employees who are likely to leave or remain in the organization. Feature importance analysis revealed that total working years, age, and overtime status significantly influenced attrition. To enhance accessibility and interactivity, the trained model was deployed through a Streamlit web application, allowing user to input attributes and obtain real-time predictions. The results reveal that positions such as age between 30-40, no overtime (30%), and marital status is divorced (25%) are the most influential features linked to attrition. This integration of predictive analytics and interactive visualization provides a practical tool for human resource decision making and proactive workforce management. The results demonstrate that machine learning can provide actionable insights to support resource strategies in retaining employees and reducing turnover risks.