Learning to Remember Patterns: Pattern Matching Memory Networks for Traffic Forecasting

This is a PyTorch implementation of the paper: Hyunwook Lee, Seungmin Jin, Hyeshin Chu, Hongkyu Lim, Sungahn Ko, Learning to Remember Patterns: Pattern Matching Memory Networks for Traffic Forecasting, ICLR 2022.

Requirements

python3
scipy>=0.19.0
numpy
pandas
pyyaml
torch>=1.9.0

Data Preparation

Download Datasets

The traffic data files for NAVER-Seoul is posted on Google Drive. The other datasets, including METR-LA, can be found in Google Drive or Baidu Yun links provided by Li et al..

Process Datasets

In the data processing stage, We have same process as Li et al.

# Create data directories
mkdir -p data/{METR-LA,PEMS-BAY,NAVER-Seoul}

# METR-LA
python generate_training_data.py --output_dir=data/METR-LA --traffic_df_fiilename=data/metr-la.h5 --seq_length_x INPUT_SEQ_LENGTH --seq_length_y PRED_SEQ_LENGTH

# PEMS-BAY
python generate_training_data.py --output_dir=data/PEMS-BAY --traffic_df_fiilename=data/pems-bay.h5 --seq_length_x INPUT_SEQ_LENGTH --seq_length_y PRED_SEQ_LENGTH

# NAVER-Seoul
python generate_training_data.py --output_dir=data/NAVER-Seoul --traffic_df_fiilename=data/naver-seoul.csv --seq_length_x INPUT_SEQ_LENGTH --seq_length_y PRED_SEQ_LENGTH

Model Training

Code and detailed instruction will be updated soon.