Official code for the KDD 2025 paper:

SDE: A Simplified and Disentangled Dependency Encoding Framework for State Space Models in Time Series Forecasting

Zixuan Weng, Jindong Han, Wenzhao Jiang, Hao Liu

Paper: https://arxiv.org/abs/2408.12068

1. Install Pytorch and necessary dependencies.

pip install -r requirements.txt

2. The datasets can be obtained from the iTransformer GitHub repository or downloaded directly from Google Drive.

3. Train and evaluate the model:

# Multivariate forecasting with SDE
bash ./scripts/SAMBA/weather.sh