<![CDATA[This study proposes a Prescriptive Maintenance (RxM) framework aimed at improving the Physical Availability (PA) of Komatsu Dump Trucks HD785-7 operated under a Full Maintenance Contract (FMC) at PT ABC Site Sangkulirang. The research integrates the DMAIC methodology with the Knowledge Discovery in Databases (KDD) process to systematically analyze operational failures. Historical breakdown data were preprocessed and modeled using a Naïve Bayes (NB) classifier, selected for its robustness in handling categorical features common in maintenance records. The model demonstrated high predictive performance with 97.93% accuracy, 100% precision, 94.12% recall, and an AUC of 0.995, indicating strong reliability in distinguishing high-risk conditions. The RxM framework was embedded into daily maintenance planning and Standard Operating Procedures (SOPs), supported by a monitoring dashboard for continuous feedback and retraining. As a result, the proportion of Breakdown Unscheduled (BUS) events decreased from 45% in 2024 to 26% in mid-2025, while fleet PA consistently exceeded the contractual target of 92%, reaching 95.5%. These findings confirm that embedding prescriptive analytics into maintenance workflows not only reduces unplanned downtime but also enhances resource allocation and decision-making. The case study highlights the practical value of combining statistical learning with structured process improvement to drive digital transformation in mining operations.]]>
