针对群体智能涌现和演化行为的预测方法在交通预测中的应用数据集

预估到达时间（ETA）任务是智能交通系统中的一项重要任务。通常，任务涉及大量的时空数据，并受到不同因素的影响，如路线距离、道路通行能力、交通信号灯和实时交通状况。实时交通状况具有高度的不确定性和动态性，这使得ETA任务具有挑战性。本课题基于SIGSPATIAL 2021 GISCUP滴滴行程时间预估任务中开放的深圳2020年8月的部分行程数据，包括出发时间、路线信息、路况信息和路网拓扑等多维度特征，依托这些数据，设计了基于地理信息注意力网络的交通状况预测和出行时间预计方法，同时预估到达时间和交通状况。数据集中包含路网数据、司机出行数据以及综合处理后的行程数据，神经网络模型数据以及部分神经网络运行结果数据。行程数据基于滴滴出行数据（https://gaia.didichuxing.com）加工获得。

Estimated Time of Arrival (ETA) task is a crucial task in intelligent transportation systems (ITS). Typically, this task involves large-scale spatiotemporal data and is influenced by multiple factors, such as route distance, road traffic capacity, traffic signals, and real-time traffic conditions. Real-time traffic conditions are highly uncertain and dynamic, which makes the ETA task particularly challenging. This study leverages partial trip data of Shenzhen in August 2020 released in the SIGSPATIAL 2021 GISCUP Didi Trip Time Estimation Challenge, which includes multi-dimensional features including departure time, route information, traffic condition data, and road network topology. Based on this dataset, we propose a traffic condition prediction and trip time estimation method based on the Geographic Information Attention Network, which simultaneously estimates the arrival time and real-time traffic conditions. The dataset contains road network data, driver travel data, comprehensively processed trip data, neural network model data, and partial neural network operation result data. The trip data was processed based on Didi Chuxing's public dataset (https://gaia.didichuxing.com).

搜集汇总

背景与挑战

背景概述

该数据集聚焦于应用群体智能涌现和演化行为的预测方法进行交通预测，其数据基础来源于滴滴出行在深圳的行程数据，并经过加工处理。数据集包含路网、司机出行、行程以及神经网络模型等多维度数据，总规模为12.81GB，由21个文件构成。