Efficient selection of trucks represents a significant challenge within logistics operations, directly influencing both cost efficiency and delivery effectiveness. This study introduces a comprehensive four-stage data modeling framework that employs machine learning techniques, including Decision Tree (DT), Random Forest (RF), and Support Vector Machine (SVM), to forecast load capacities and appropriately assign trucks for deliveries. The analysis utilized historical delivery data, which was meticulously processed to ensure accurate predictions regarding truck selection based on load capacity. Findings indicate that the Random Forest model excels in precisely estimating pallet requirements, while the Decision Tree method demonstrates superior performance in optimizing truck assignments. This research provides logistics managers with a viable strategy to enhance operational efficiency, reduce expenses, and improve resource allocation.