This repository contains a PyTorch implementation of the MSTDP presented in "Where to Go Next Day: Multi-Scale Spatial-Temporal Model for Mid-term Human Mobility Prediction"
- python==3.8.5
- pytorch==1.13.1
- torch_geometric==2.3.1
- numpy
- pandas
- geopandas
- shapely
- scipy
- tqdm
- data: file directory for datasets.
- models: file directory to define the model and its modules.
- results: file directory to store experimental results.
- utils: file directory with various tools.
- data_prepare.py: used to prepare the trajectory and graph datasets.
- main.py: the entrance to the main process.
It is recommended to adaptively modify the code based on the specific dataset employed, in order to accomplish the construction of trajectories and graphs.
python data_prepare.py --[Parameter set]
A detailed description of the parameters is provided in .\utils\parameter_settings.py. It is advised to adaptively modify parameters when applying to a new dataset, such as traj_len, lid_size, tract_size, etc.
python main.py --[Parameter set]