Abstract:
Student mental health is an important aspect of overall student well-being, but it is a less explored area with respect to the Bengali participants. We are using machine learning methods to predict the mental health of Bengali students and this study is also reflects on the factors that affect students’ mental health. Using data from student responses, we trained and compared four machine learning models: Logistic Regression, Random Forest, XGBoost, and Support Vector Classifier (SVC). Each model was evaluated based on performance metrics like accuracy, precision, recall, and F1 score. Logistic Regression proved to be the best model out of all of them, with the highest accuracy (94%) and well-balanced metrics; 86% precision, 100% recall, 92% F1 score. Both Random Forest and SVC models showed competitive performance, providing more descriptive analysis about the mental health classification, while XGBoost slightly lagged behind but maintained a balanced score. Thus, these results could shine a light on how data can be used to identify and mitigate mental health issues in educational settings. While the study is promising regarding the potential of machine learning in predicting mental health, the researchers acknowledged limitations such as dataset surround and survey potential bias. Further studies could increase sample size, improve survey measures for sociological parameters, and create monitoring systems that aim to increase prediction accuracy in real-time. This work establishes a groundwork for early mental health interventions and provides educators, guardians, and mental health professionals tools to promote the well-being of Bengali students. Through the integration of data-driven insights into educational structures, the objective of this research is to promote early detection and intervention strategies for mental health challenges in academic settings
