Answers are grounded in this paper and cite their sources.
Observational analysis shows random forest regression improves delivery time predictions, indicating better logistics efficiency in urban areas.
Predicting delivery times is a central challenge in logistics, especially during the last-mile phase, which is the final step in the supply chain where goods are delivered to the customer. This work studies previous approaches and proposes Random Forest Regression for Time Prediction (hereafter RFRTP), a Random Forest Regression-based model to improve delivery time predictions in logistics systems. This study examines the effectiveness of RFRTP compared to baseline statistical and machine learning models when accounting for the complex factors affecting delivery times, such as traffic conditions, order density, and route assignments. A dataset based on delivery data provided by JD Logistics, containing extensive information on last-minute deliveries in an urban environment, was used to train and test RFRTP and multiple baseline regression models, specifically Linear, Lasso, and Ridge, which were analyzed on their performances in delivery time prediction. The experimental results show that the proposed RFRTP model obtains the best results in last-minute delivery time prediction, which strongly contributes to more reliable time estimates, improved customer satisfaction, and greater efficiency. Through feature importance analysis on the RFRTP model, this study found the most critical dataset features for prediction, providing insight into the most important data. This study also calculated prediction intervals for the results of RFRTP, allowing for margins of error and transparency in the confidence of the models estimates.
No takes yet. Share an insight, caveat, or question.
Allen He (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: