We present pytorch implementation of the popular deep-learning models in the following git repository:
Github: https://github.com/cdsnlab/demand-pytorch
demand-pytorch
Results
We report MAE in Beijing taxi flow prediction dataset for each prediction steps.
| Model | bj-flow (15 min) | bj-flow (30 min) | bj-flow (1 hour) |
|---|---|---|---|
| ConvLSTM | |||
| STResNet | |||
| DeepSTN |
We report RMSE and MAPE in NYC taxi flow prediction dataset.
| Model | RMSE | MAPE |
|---|---|---|
| DMVST-Net |
Getting Started
Data
- Beijing Taxi: download bj-flow.pickle from Google Drive
- NYC Taxi:
Environment
conda create -n $ENV_NAME$ python=3.7
conda activate $ENV_NAME$
# CUDA 11.3
pip install torch==1.11.0+cu113 --extra-index-url https://download.pytorch.org/whl/cu113
# Or, CUDA 10.2
pip install torch==1.11.0+cu102 --extra-index-url https://download.pytorch.org/whl/cu102
pip install -r requirements.txt
Train
If config file not specified, load $MODEL_NAME$_config.py by default.
python train.py --model $MODEL_NAME$ --ddir $PATH_TO_DATASET$ --dname $DATASET_NAME$