<![CDATA[Optimal management of State-Owned Assets (BMN) is a crucial factor in ensuring the operational effectiveness of government institutions, particularly at the District Attorney's Office of Tangerang Regency. However, unexpected asset damage often disrupts workflows and leads to inefficiencies in maintenance budgets. This study addresses this issue using a data mining approach, with the primary objective of evaluating and comparing the performance of three machine learning algorithms: K-Nearest Neighbor (K-NN), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost) in predicting asset damage. The methodology employs a classification experimental study on historical asset maintenance data from the 2021–2025 period, analyzed using the Orange Data Mining platform. The research process includes data preprocessing stages (imputation and normalization), data partitioning into 70% training and 30% testing sets, and performance evaluation based on Area Under Curve (AUC), Accuracy, F1-Score, Precision, and Recall metrics. Comparative analysis results indicate that the XGBoost algorithm delivers superior performance, achieving the highest AUC of 0.987 and an F1-Score of 0.905, along with dominant prediction accuracy compared to other models. The K-NN algorithm demonstrates good and stable performance in the second position, whereas SVM exhibits lower accuracy levels than the other two models. Based on these findings, XGBoost is recommended as the optimal model for implementation within the District Attorney's Office asset management system. The adoption of this model is expected to support strategic decision-making, enable a transition to predictive maintenance, and enhance the efficiency of state asset management.]]>
