Grid and Graph-based Autonomous Transportation System State Awareness Open Dataset

**I. Dataset Overview** This dataset is based on 7 days of connected vehicle trajectory data collected from part of the road network in Jiading District. It provides an open dataset for global state awareness of autonomous transportation systems based on both grid and graph structures. The dataset includes multiple types of information, such as traffic flow and average vehicle speed, and can be used for research on traffic flow prediction, traffic pattern recognition, route optimization, spatiotemporal analysis, and related topics. **II. Dataset Contents** This dataset contains both grid-based and graph-based data for Jiading District. It records the traffic flow and average vehicle speed of each grid and road node at each time step. The time span is from April 15 to April 21, 2024, totaling 7 days. The grid data divides Jiading District into 1 km × 1 km grids. The road network data of Jiading District was obtained from OpenStreetMap, based on which graph node IDs were assigned. The data sampling frequency is 30 minutes. In particular, for privacy protection, the dataset does not contain specific vehicle IDs. **III. Dataset Files** The dataset is provided as **jiading_data.zip**. The archive contains two folders: **ShanghaiGraph** and **ShanghaiGrid**, which store the graph-based and grid-based traffic data, respectively. **IV. Data File Description** The file structure of the dataset is as follows: jiading_data.zip ShanghaiGraph/|-- graph_data.npz # Graph data|-- ShanghaiGraph.csv # Spatial adjacency relationships of the graph structure; the "cost" field represents the actual distance (m) ShanghaiGrid/|-- grid_data.npz # Grid data|-- ShanghaiGrid.csv # Spatial adjacency relationships of the grid structure; the "cost" field represents the actual distance (m) The fields contained in **graph_data.npz** and their meanings are as follows: ['data', 'data_nxtc', 'time_slots', 'node_count', 'time_steps', 'feature_names'] | Column No. | Type | Description ||---|---|---|| 1 | data_nxtc | Main data with shape (N, T, C), where N is the number of grids, T is the number of consecutive time steps, and C is the feature dimension. It is typically used for time-series prediction tasks. || 2 | data | Same as the main data. || 3 | time_slots | Time slot information, indicating the specific time of each data point. || 4 | node_count | Number of nodes, indicating the number of nodes in the graph. || 5 | time_steps | Number of time steps, indicating the temporal dimension of the data. || 7 | feature_names | Feature names, listing all features in the dataset and indicating the specific meaning of each data column. | The fields contained in **grid_data.npz** and their meanings are as follows: ['data_nxtc', 'data', 'dates', 'node_count', 'time_steps', 'day_count', 'feature_names'] | Column No. | Type | Description ||---|---|---|| 1 | data_nxtc | Main data with shape (N, T, C), where N is the number of nodes, T is the number of consecutive time steps, and C is the feature dimension. It is typically used for time-series prediction tasks. || 2 | data | Data split by date. || 3 | dates | Date information, indicating the specific date of each data entry. || 4 | node_count | Number of nodes, indicating the number of nodes in the graph. || 5 | time_steps | Number of time steps, indicating the number of time steps in the data. || 6 | day_count | Number of days, indicating the date range of the dataset. || 7 | feature_names | Feature names, listing the names of all node features in the graph data. | **V. Data Statistics** **Graph Dataset** Number of nodes: 2,752 Average missing rate: 42.22% Time interval: 30 minutes Traffic flow statistics: mean = 66.86, p50 = 18, p95 = 254, max = 5,239 (non-zero positions) Average speed statistics: mean = 38.99, p50 = 30.75, p95 = 90.67, max = 125.1 (non-zero positions) Average traffic flow per node per time interval: 38.63 (including missing values) **Grid Dataset** Grid size: 1 km × 1 km Number of grids: 624 Average missing rate: 47.9% Time interval: 30 minutes Traffic flow statistics: mean = 121.29, p50 = 37, p95 = 541, max = 2,597 (non-zero positions) Average speed statistics: mean = 37.27, p50 = 31.12, p95 = 83.38, max = 127 (non-zero positions) Average traffic flow per grid per time interval: 63.1971 **Data Provider** This dataset is provided by Beijing University of Posts and Telecommunications and is funded by the National Key R&D Program of China under the project “Computational Technologies for Autonomous Transportation Systems (2023YFB4301900).”