AI 中文总结
本研究针对城市食品配送的交付时间预测问题，评估多种机器学习模型，发现XGBoost性能最优，可支持低碳配送决策与优化。
AI 中文摘要
在线食品配送服务需求的日益增长显著影响了城市物流，带来了与运营效率和环境可持续性相关的关键挑战。在此背景下，准确的交付时间预测在提高服务质量的同时，支持向低碳配送系统的转型中发挥着关键作用。本文首先对城市食品配送的可持续和低碳配送模式进行了全面分析。随后，研究开发并评估了多种机器学习模型，即XGBoost、LightGBM、梯度提升（Gradient Boosting）和KNN，使用结合了运营、时间和上下文特征的数据集来预测食品配送时间。比较分析表明，基于树的模型优于其他模型，其中XGBoost取得了最佳的预测性能。研究结果证明了人工智能和机器学习技术在提高配送效率和支持可持续城市物流决策方面的巨大潜力。此外，所提出的预测框架可以集成到优化模型中，以增强配送规划并支持选择环境可持续的配送模式。
英文摘要
The increasing demand for online food delivery services has significantly impacted urban logistics, raising critical challenges related to operational efficiency and environmental sustainability. In this context, accurate delivery time prediction plays a key role in improving service quality while supporting the transition toward low-carbon delivery systems. This paper first provides a comprehensive analysis of sustainable and low-carbon delivery modes for urban food distribution. The study then develops and evaluates several machine learning models, namely XGBoost, LightGBM, Gradient Boosting, and KNN, using a dataset that combines operational, temporal, and contextual features to predict food delivery times. The comparative analysis shows that tree-based approaches outperform the other models, with XGBoost achieving the best predictive performance. The findings demonstrate the strong potential of artificial intelligence and machine learning techniques for enhancing delivery efficiency and supporting sustainable urban logistics decision-making. Moreover, the proposed predictive framework can be integrated into optimization models to enhance delivery planning and support the selection of environmentally sustainable delivery modes.
