Optimasi Model XGBoost dengan Genetic Algorithm untuk Prediksi Kesehatan Mental Siswa Sekolah Menengah Berbasis Machine Learning

Abstract

Mental health is a vital aspect of human well-being, yet often neglected. Recent studies report a rise in depression, anxiety, and stress among adolescents, especially post-COVID-19. Machine learning has emerged as a powerful tool for predicting mental health conditions. This study employs the XGBoost Regressor using a regression-based ML approach to predict mental health high school students. To enhance accuracy, hyperparameter optimization is conducted using a Genetic Algorithm (GA) to identify the optimal parameter set. The baseline model achieved an MSE of 0.3698, RMSE of 0.6081, and MAPE of 14.09%. After GA optimization, performance improved to an MSE of 0.3092 (16.4% reduction), RMSE of 0.5560 (8.6% reduction), and MAPE of 12.88% (8.6% reduction). These results demonstrate the model's effectiveness for early mental health screening in educational settings, enabling timely interventions by school counselors and healthcare providers.