Austin R. Benson, David F. Gleich, and Lek-Heng Lim

This repository contains research code to accompany our work:

The Spacey Random Walk: a Stochastic Process for Higher-order Data.

Available on arXiv.

The synthetic data used in the numerical experiments is also included in the repository in processed_data/synthetic_experiments/. The data can also be generated again:

  make sim
  bash generate_synthetic.sh

To train and test the spacey random walk model on the data, run:

  make learn
  bash learn_synthetic_R1_R2.sh
  bash learn_synthetic_random.sh

The processed taxi trajectory data used in the numerical experiments is provided in the file processed_data/taxi/manhattan-year-seqs.txt. The raw data can be downloaded here. The code used to process this data is in scripts/form_taxi_trajectories.py.

To train and test the spacey random walk model on the data, run:

  make learn
  bash learn_taxi.sh