This is the repo for the ITSC 2022 paper "Forecasting Regional Multimodal Transportation Demand with Graph Neural Networks: An Open Dataset" , the deep learning dataset and research in NYC taxi and bike prediction.
The original data is from the public datasets of NYC taxi and bike. The datasets can be downloaded from the websites below.
Bike: Citi Bike System Data | Citi Bike NYC
Taxi: TLC Trip Record Data - TLC (nyc.gov)
And we have done some data pre-processing for the Bike and Taxi datasets.
| Graph-Based | Data Description / Data Source | Spatial Domain | Time Period | Time Interval |
|---|---|---|---|---|
| METR_LA | Traffic Speed Sensors in Los Angeles CountyLos Angeles Metropolitan Transportation Authority*Collaborated with University of Southern California https://imsc.usc.edu/platforms/transdec/ | 207 sensors | 2012/3/1∼2012/6/30 | 60 minutes |
| BikeNYC | Bike In-Out Flow / Bike Trip Data of New York City https://www.citibikenyc.com/system-data | 69 regions | 2019/1/1~2020/12/31 | 30/60 minutes |
| TaxiNYC | Taxi In-Out Flow / Taxi Trip Data of New York City The New York City Taxi&Limousine Commission (TLC) https://www1.nyc.gov/site/tlc/about/data.page | 69 regions | 2019/1/1~2020/12/31 | 30/60 minutes |
The project structure is given as below:
- data-NYCBike: The npz file for NYCBike datasets
- data-NYCTaxi: The npz file for NYCTaxi datasets
- data-NYCZones: The zones information for New York and the adjacency matrix
- data-processing: Generate the npy and npz files from the origin data
- model: Model and prediction for NYC data with STGCN, DCRNN and Graph-waveNet
- visualization: The prediction result for STGCN, DCRNN and Graph-waveNet, and some visualization