Predictive Modelling of Mental Health Disorders Using Machine Learning Classifiers

Student mental health problems have been a major concern in recent years, having an effect on both academic achievement and general well-being. An increase in stress, anxiety, and other mental health illnesses has been attributed to the pressures of academic life, societal expectations, and personal challenges. Through the use of an extended dataset, improved feature engineering, and the application of complex machine learning models, this project seeks to improve predictive analytics for student mental health. Our dataset comprises of almost 20,000 entries with features like age, gender, family income, family size, CGPA and academic pressure. In order to forecast student's stress levels, we assessed a variety of predictive models, including Logistic Regression, SVM, Random Forest, KNN, Decision tree and Gradient Boosting. We have also compared these techniques in which Random Forest, Decision tree and Gradient Boosting models demonstrated superior classification performance. The study also highlights the moral issues involved in using mental health data for forecasting. The findings show increased interpretability and accuracy across models, offering insightful information for creating early intervention plans that effectively address mental health issues in educational environments.