Next location prediction using transformers

This repository represents the implementation of the paper:

How do you go where? Improving next location prediction by learning travel mode information using transformers

Ye Hong, Henry Martin, Martin Raubal
| IKG, ETH Zurich | IARAI |

Requirements and dependencies

This code has been tested on

• Python 3.9.12, Geopandas 0.12.2, trackintel 1.1.13, PyTorch 1.12.1, cudatoolkit 11.3, GeForce RTX 3090

To create a virtual environment and install the required dependences please run:

    git clone https://github.com/mie-lab/location-mode-prediction
    cd location-mode-prediction
    conda env create -f environment.yml
    conda activate loc-mode-pred

in your working folder.

Hyperparameters

All hyperparameter settings are saved in the .yml files under the respective dataset folder under config/. For example, config/geolife/transformer.yml contains hyperparameter settings of the transformer model for the geolife dataset.

Reproducing on the Geolife dataset

The results in the paper are obtained from SBB Green Class and Yumuv dataset that are not publicly available. We provide a runnable example of the pipeline on the Geolife dataset. The travel mode of the Geolife users are determined through the provided mode labels and using the trackintel function trackintel.analysis.labelling.predict_transport_mode. The steps to run the pipeline are as follows:

1. Install dependencies

• Download the repo, install neccessary Requirements and dependencies.

2. Download Geolife

• Download the Geolife GPS tracking dataset from here. Unzip and copy the Data folder into geolife/. The file structure should look like geolife/Data/000/....
• Create file paths.json, and define your working directories by writing:

{
    "raw_geolife": "./geolife"
}

3. Preprocess the dataset

• run

    python prePro/geolife.py geolife 20

for executing the preprocessing script for the geolife dataset. The process takes 15-30min. locations_geolife.csv and dataset_geolife.csv will be created under data/ folder.

4. Run the proposed transformer model

• run

    python main.py config/geolife/transformer.yml

for starting the training process. The dataloader will create intermediate data files and saves them under data/temp/ folder. The configuration of the current run, the network paramters and the performance indicators will be stored under outputs/ folder.

Citation

If you find this code useful for your work or use it in your project, please consider citing:

@inproceedings{Hong_2022_how,
    author      = {Hong, Ye and Martin, Henry and Raubal, Martin},
    title       = {How do you go where? Improving next location prediction by learning travel mode information using transformers},
    year        = {2022},
    doi         = {10.1145/3557915.3560996},
    booktitle   = {Proceedings of the 30th International Conference on Advances in Geographic Information Systems (SIGSPATIAL '22)},
    location    = {Seattle, Washington, USA},
    publisher   = {Association for Computing Machinery},
    address     = {New York, NY, USA},
}

Contact

If you have any questions, please open an issue or let me know:

• Ye Hong {hongy@ethz.ch}