COMPARATIVE STUDY OF MULTINOMIAL LOGISTIC REGRESSION AND RANDOM FOREST ALGORITHMS FOR PREDICTING PSYCHOLOGICAL WELLNESS AMONG STUDENTS MENTAL HEALTH SURVEY

Stress represents a common emotional reaction to stimulating conditions however, when it becomes excessive particularly among students it may result in serious mental health syndromes, physical uneasiness, self-harm. This research aims to classify student wellness levels by inspecting a broad range of stress-related proportions, including psychological, physiological, academic, social, and ecological aspects. This study investigates student psychological wellness using machine learning techniques. A mental health survey dataset from Kaggle was analysed to predict wellness levels based on stress, anxiety, depression, sleep, diet, physical activity, and social factors. A Multinomial Logistic Regression (MLR) model and a Random Forest (RF) model were applied to the data. A Wellness Level was calculated to classify students into Low, Average, Moderate, and High wellness categories. Performance of both models was evaluated using accuracy, precision, recall, F1-score, and AUC-ROC. Results indicated that the MLR model achieved slightly higher predictive accuracy than RF. The findings suggest that early mental health screening can be effectively supported by predictive models. This contributes to proactive student care through targeted interventions. The models help institutions allocate resources more efficiently. Overall, machine learning presents a valuable approach for psychological wellness prediction. The study promotes data-driven decisions in academic mental health strategies.