Readme

This repository is a fork from the BasicTS framework (https://github.com/GestaltCogTeam/BasicTS) proposed in the work Exploring Progress in Multivariate Time Series Forecasting: Comprehensive Benchmarking and Heterogeneity Analysis (https://arxiv.org/abs/2310.06119)

Besides some utility edits, two new architectures have been added:

• STDMAE + DCRNN
• STDMAE + D2STGNN

This was done as part of my martes's thesis Prediction of traffic flow levels with Graph Neural Networks

💿 Set up

A Linux environment is recommended (e.g., Ubuntu or CentOS). We recommend using Anaconda to create a virtual Python environment.

# Install Python
conda create -n BasicTS python=3.11
conda activate BasicTS
# Install PyTorch
pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu121
# Install other dependencies
pip install -r requirements.txt

In case of problems, refer to the original_requirements.txt file

You can download the all_data.zip file from Google Drive or Baidu Netdisk. Unzip the files to the datasets/ directory:

cd /path/to/BasicTS # not BasicTS/basicts
unzip /path/to/all_data.zip -d datasets/

In order to use the STD-MAE pretraining technique, some extra libraries are required:

pip install positional_encodings==6.0.1
pip install timm==0.6.7

Download the weights of the two autoencoders from the official repository:

# Download
wget https://raw.githubusercontent.com/Jimmy-7664/STD-MAE/main/mask_save/SMAE_PEMS03_864.pt
wget https://raw.githubusercontent.com/Jimmy-7664/STD-MAE/main/mask_save/TMAE_PEMS03_864.pt

# Positioning
mv /SMAE_PEMS03_864.pt /BasicTS/baselines/STDMAE/mask_save/
mv /TMAE_PEMS03_864.pt /BasicTS/baselines/STDMAE/mask_save/

Run the demo

One the environment is set up, you can run the web app with

python3 demo/web_app.py