<![CDATA[Rural and Urban Land and Building Tax (PBB-P2) is an important source of Regional Original Income (PAD). However, suboptimal taxpayer compliance often results in under-targeted regional tax revenues. Conventional taxpayer compliance management encourages the need for Artificial Intelligence (AI) and machine learning-based approaches to support more proactive, data-driven decision-making. This study uses the Extreme Gradient Boosting (XGBoost) algorithm to predict PBB-P2 taxpayer compliance based on historical tax data. The research steps include data preprocessing, splitting the training and testing datasets with a 70:30 ratio, model training, hyperparameter tuning, and model evaluation using precision, recall, F1-score, and confusion matrix. The results showed that the best model produced an F1-Macro value of 0.7108 with a learning rate of 0.2, max depth of 12, n_estimators of 400, min_child_weight of 1, subsample of 0.8, and gamma of 0. The most influential variables in the prediction included the District Code, Principal Payment Amount, and Village Code. The XGBoost model was able to provide quite good classification performance in supporting the identification of PBB-P2 taxpayer compliance more effectively.]]>
Copyrights © 2026
