PREDICTION OF DEPRESSION AMONG NIGERIA STUDENTS USING MACHINE LEARNING: A CASE STUDY OF JOSEPH AYO BABALOLA UNIVERSITY

Student depression, often undetected due to its complexity, threatens well-being and academic success. Accurate prediction enables early intervention, mental health support, and policy improvements. However, existing methods often lack robustness and generalizability. To address this, I developed a Python-based prediction pipeline using Scikit-learn, comparing Logistic Regression, Random Forest, and Gradient Boosting. The dataset included 27,898 student samples (58.6% depressed) with 109 features. Models were evaluated via stratified 5-fold cross-validation on Accuracy, Precision, Recall, F1-Score, and ROC AUC. Logistic Regression performed best, achieving: Accuracy: 0.8481 ± 0.0018, Precision: 0.8580, Recall: 0.8874, F1-Score: 0.8725, ROC AUC: 0.9213, These results demonstrate strong predictive capability, offering a practical tool for proactive mental health initiatives. Even established algorithms can deliver meaningful impact in public health.