README

This is the official implementation of TimeCapsule presented in the paper "TimeCapsule: Solving the Jigsaw Puzzle of Long-Term Time Series Forecasting with Compressed Predictive Representations" https://arxiv.org/abs/2504.12721.

🎉🎉Our work has been accepted by KDD 2025.

Introduction

TL;DR See our oral slides for detailed motivations and explanations regarding the model. slides

Get Started

1. Please download the dataset you need at Google Drive or Baidu Cloud.

2. Note the position you place the dataset package and modify the path in scripts files accordingly.

3. Change to the correct path, then use bash to run the specific script file, e.g., bash ./scripts/traffic.sh.

Acknowledgment

1. We recognize that this is an imperfect work. Please let us know if you have any questions or suggestions about the code or specific techniques in the paper.
2. It is recommended to increase the length of compression parameters when using short look-back windows, other hyperparameters should also be adjusted accordingly.
3. Please consider starring this repository and citing our research if you find the paper interesting or this repo helpful.

@article{lu2025timecapsule,
  title={TimeCapsule: Solving the Jigsaw Puzzle of Long-Term Time Series Forecasting with Compressed Predictive Representations},
  author={Lu, Yihang and Xu, Yangyang and Qing, Qitao and Meng, Xianwei},
  journal={arXiv preprint arXiv:2504.12721},
  year={2025}
}

We appreciate the following repos for their valuable code and efforts.

• PatchTST
• iTransformer