Classification of Depression Severity Using a Random Forest Algorithm Based on Lifestyle, Demographic, and Psychological Factors

Abstract

Depression among college students is a mental health issue that impacts quality of life and academic performance. However, factors influencing depression levels such as lifestyle, demographics, and psychological factors have not yet been analyzed in an integrated manner. This study aims to develop a depression severity classification model using the Random Forest algorithm based on these factors. The dataset consists of 1,998 records with 16 features selected through the Knowledge Discovery in Database (KDD) process. To address data imbalance, the Synthetic Minority Over-sampling Technique (SMOTE) was applied. The results show that the Random Forest model achieved an accuracy of 97.88% and an AUC of 0.998. Feature importance analysis indicates that the variables Symptoms, Nervous Level, and Employment Status are dominant factors in determining depression levels. Based on these results, the model is capable of effectively classifying depression levels and has the potential to serve as the basis for an early detection system in the university setting.