<![CDATA[Abstract−Mental health problems among adolescents have emerged as a serious concern in global scientific discourse, particularly as digital media usage intensifies. This study aims to develop a depression risk classification model for adolescents aged 13–19 years using the C4.5 (J48) Decision Tree algorithm, with predictor variables including daily social media usage duration, sleep quality and quantity, academic performance, and stress level. The data source is the Teen Mental Health Dataset from Kaggle, comprising 1,200 records with 13 attributes collected from 2022 to 2024. All analytical stages—from data preprocessing to model evaluation—were conducted using RapidMiner Studio. Model validation was performed through 10-fold cross-validation, yielding accuracy of 92.5%, precision of 91.8%, recall of 89.3%, and F1-Score of 90.5%. Based on feature importance analysis, stress_level, anxiety_level, and addiction_level proved to be the three dominant predictors of depression risk. TikTok users consistently showed higher stress levels than Instagram or combined platform users. The resulting model is interpretable and has potential for direct integration into school counseling practice and digital health research.]]>
Copyrights © 2026
