Machine Learning-Based Prediction of the Impact of Mental Health Policies on Employee Productivity

Abstract

This research looks at the relationship between mental health, work-life integration, and employee productivity accompanied by stress level, work-life conflict, and the availability of mental wellbeing resources. Using two suitable machine learning techniques, these are Gradient Boosting and Random Forest, this research assesses the extent of change in productivity as a result of these three factors. The analysis revealed the existence of a relationship between the three variables Stress Level, Work Life Balance Rating, and Access to Mental Health Resources and productivity levels among the employees. The models performed similarly with respect to Mean Absolute Error (MAE), R-squared (R²) values on both models but there were challenges in predicting productivity classes due to low AUC value. Recommendations follow from these findings suggesting company policies which should mandate the need for further improvement on workplace mental health through appropriate stress management, working arrangements, and mental wellness programs. This study provides data enhancement and application of sophisticated predictive models techniques which should be employed in further studies to boost prediction precision.