Digital Twin Mobility Profiling: A Spatio-Temporal Graph Learning Approach

This is the implementation of traffic prediction code in DTMP based on PyTorch.

structure of the code:

• data folder: storing PEMSD4 and PEMSD8 dataset. You may refer to ASTGCN for more details;
• lib folder: some methods for data loading and processing from AGCRN;
• utils.py: method of loading adjacency graph;
• model.py: implementation of Anet;
• train.py, run.py: train and run the model.

You can use python run.py --dataset PEMSD4 --num_nodes 307 command to run the code.