Predicting Student Stress Levels Based on Lifestyle Factors Using the Catbost Algorithm

<table width="593" border="0" cellspacing="0" cellpadding="0"><tbody><tr><td valign="top" width="387"><p>This study developed a machine learning model to classify student stress levels based on lifestyle factors using the CatBoost algorithm. Data were collected from 630 students of the SciTech Faculty at State Islamic University of North Sumatra through a questionnaire comprising 14 Likert-scale items. Instrument validation was confirmed using Pearson’s r (>0.821, p < 0.05) and Cronbach’s Alpha (0.866). Preprocessing included outlier removal with IQR, feature encoding, stratified train-test split (80:20), and 5-fold cross-validation. The training set was imbalanced and addressed using the SMOTE technique. Model evaluation used accuracy (85%), precision, recall, and F1-score per class, with high recall (0.97) for moderate and improved F1-score (0.79) for low stress. Final classification used a 20% test subset (126 samples). Feature importance analysis identified task procrastination, poor sleep quality, and weak time management as key predictors. These findings affirm CatBoost's reliability through consistent results, scalability, and balanced evaluation metrics beyond mere accuracy.</p></td></tr></tbody></table>