This study employs advanced machine learning techniques—Random Forest and Logistic Regression—to enhance the prediction and analysis of delivery speeds in urban freight logistics. By integrating a comprehensive dataset encompassing variables such as route distance, traffic conditions, driver demographics, and vehicle characteristics, we provide a nuanced exploration of the factors influencing delivery speeds. Our findings reveal significant predictors including traffic speed and driver age, challenging traditional assumptions about peak traffic impacts. The Random Forest model excels in handling complex, non-linear interactions among factors, while Logistic Regression offers insights into the direct influences on delivery outcomes. This research contributes to urban planning and logistics by offering empirically backed, actionable insights for optimizing delivery routes and schedules, thereby improving urban mobility and reducing environmental impacts. These outcomes support sustainable urban development by facilitating more efficient and predictive urban freight logistics operations.