Violence against children is a social issue that affects children's physical, psychological, and social development. This study aims to classify the level of child violence risk based on household socio-economic profiles in Jember Regency using the CatBoost algorithm. The study employed a quantitative non-experimental approach using socio-economic data from 31 districts, which underwent preprocessing stages including data cleaning and transformation. Risk labeling was performed using the 33–66 quantile method, while model validation was conducted using Leave-One-Out Cross Validation (LOOCV). The results showed that the CatBoost model achieved an accuracy of 97% and a ROC-AUC score of 0.991. In addition, the average validation accuracy was 0.968 ± 0.177. The classification results indicated that 12 districts were categorized as high risk, 9 districts as medium risk, and 10 districts as low risk. SHAP analysis revealed that education level and type of occupation were the dominant variables contributing to the classification results. This study demonstrates that CatBoost is capable of providing strong classification performance in identifying child violence risk based on socioeconomic data.
Copyrights © 2026