To successfully run the code, you need the following dependencies:
- pandas 1.1.4
- numpy 1.19.4
- scikit-mobility
- tqdm
- textdistance
The data folder contains the dataset of Foursquare NYC already processed. Due to the significant weights of the files and the limits for the supplementary material, we only provide one of the datasets presented in the main paper. However, in the code folder, we also provide two notebooks, namely data_downloader.ipynb and data_preparation.ipynb, to let you download and preprocess the other datasets.
The script overlap.py contains the logic to compute the overlap of the test trajectories and produce the train-test stratification. The new computed files are automatically saved in the data folder with the names dataset_metric_overlap.pkl.
Please note that running overlap.py may take a while.
The files are in a convenient format and can be directly used as input to well-known mobility models like Deep Move https://github.com/vonfeng/DeepMove