NODA

This repository is the official implementation of NODA in ''Neighbor Geospatial Self-Attention for Crowd Flow Prediction''

Appendix (PDF) : The appendix include A) Related Work, B) Definitions, C) Algorithm, and D) Outline of the Baselines

Requirements

To install requirements:

pip install -r requirements.txt

Experiment

1. Download data from here and put into /data/NYC or /data/DC.

2. We provide the experiment scripts of all models under the folder ./scripts. You can reproduce the experiment results by:

   ./scripts/Main/NYC/NODA.sh
   ./scripts/Main/DC/NODA.sh
   ./scripts/Ablation/Transformer.sh
   ./scripts/Ablation/Attention.sh
   

Acknowledgement

We appreciate the following github repo a lot for their valuable code base

https://github.com/thuml/Autoformer

https://github.com/jonpappalord/crowd_flow_prediction

The dataset is provided by:

https://capitalbikeshare.com/system-data

https://citibikenyc.com/system-data