Accurate point and distributional forecasts across diverse horizons are crucial for time-series forecasting. However, existing research often focuses on isolated aspects, such as long-term point forecasting or short-term probabilistic estimation. This raises a fundamental question: How do different methodological designs address these diverse forecasting needs?
In this repository, we:
- Provide Detailed Reproduction Guides: Offer comprehensive instructions for replicating supervised models and pre-trained foundation models.
- Evaluate Methods Under a Unified Framework: Align and assess existing methods across various data scenarios using a consistent benchmarking framework.
- Deliver In-Depth Insights: Present detailed analyses and insights into the experimental results.
