Last-mile delivery route deviations dataset: planned vs. actual routes

Logistics companies use specialized optimization algorithms to plan delivery routes that minimize costs and distances. However, drivers frequently deviate from these planned routes based on their knowledge.

Dataset Overview: The dataset tracks last-mile delivery routes across two countries for a logistics company, given the planned stop sequence generated by routing software with the actual sequence followed by drivers. Each route also includes driver identification, arrival information, time windows, and distances between stops in planned and actual routes. Multiple stops may share the same location. Time values are encoded as differences from the earliest timestep in each route. Distances between stops are calculated using the Haversine formula.

物流企业采用专业化优化算法规划配送路线，以最小化运营成本与行驶里程。但驾驶员常会凭借自身实操经验偏离预设的配送路线。

数据集概览：本数据集记录了某物流公司覆盖两个国家的末端配送（last-mile delivery）路线，包含路由规划软件（routing software）生成的预设停靠点序列，以及驾驶员实际执行的停靠序列。每条路线还附带驾驶员标识、到达信息、时间窗（time windows），以及预设与实际路线中各停靠点间的距离。多个停靠点可共享同一地理位置。时间值以每条路线内最早时间步的时间差进行编码。停靠点间的距离通过哈弗辛公式（Haversine formula）计算得出。