Multi-location Traffic Flow Dataset: Stockton and Oakland California

This dataset supports the research detailed in "A Hybrid Deep Learning Approach with GEH-based Loss Function and Evaluation Metric for Multi-location Traffic Flow Forecasting" containing traffic flow measurements from Stockton and Oakland, California. The dataset is organized into four sub-folders: 'Oakland-raw', 'Oakland-processed', 'Stockton-raw', and 'Stockton-processed'. The raw data folders ('Oakland-raw' and 'Stockton-raw') include the initial traffic measurements as collected directly from traffic sensors, providing a granular view of traffic patterns without any modifications. The processed data folders ('Oakland-processed' and 'Stockton-processed') contain data that has been cleaned and structured, making it more suitable for analytical tasks and modeling purposes.

本数据集支持发表于《面向多区域交通流预测的基于GEH损失函数与评价指标的混合深度学习方法》的相关研究，包含美国加利福尼亚州斯托克顿（Stockton）与奥克兰（Oakland）的交通流量实测数据。该数据集分为四个子文件夹：'Oakland-raw'、'Oakland-processed'、'Stockton-raw'以及'Stockton-processed'。其中原始数据文件夹（'Oakland-raw'与'Stockton-raw'）收录了直接从交通传感器采集的初始交通流量实测数据，未经过任何修改，可完整呈现交通模式的细粒度特征。经处理的数据文件夹（'Oakland-processed'与'Stockton-processed'）则包含经过清洗与结构化处理的数据，更适配各类分析任务与建模需求。