- 23-01-2025: 🚀 Submitted to GIFT-EVAL benchmark, stay tuned for results!
- 10-10-2024: 🚀 TabPFN-TS paper accepted to NeurIPS 2024 TRL and TSALM workshops!
We demonstrate that the tabular foundation model TabPFN, when paired with minimal featurization, can perform zero-shot time series forecasting. Its performance on point forecasting matches or even slightly outperforms state-of-the-art methods.
Our work proposes to frame univariate time series forecasting as a tabular regression problem.
Concretely, we:
- Transform a time series into a table
- Extract features from timestamp and add them to the table
- Perform regression on the table using TabPFN
- Use regression results as time series forecasting outputs
For more details, please refer to our paper and our poster (presented at NeurIPS 2024 TRL and TSALM workshops).
- Zero-shot forecasting: this method is extremely fast and requires no training, making it highly accessible for experimenting with your own problems.
- Point and probabilistic forecasting: it provides accurate point forecasts as well as probabilistic forecasts.
- Support for exogenous variables: if you have exogenous variables, this method can seemlessly incorporate them into the forecasting model.
On top of that, thanks to tabpfn-client from Prior Labs, you won’t even need your own GPU to run fast inference with TabPFN. 😉 We have included tabpfn-client as the default engine in our implementation.
The demo should explain it all. 😉
We have submitted our results to the GIFT-EVAL benchmark. Stay tuned for results!
For more details regarding the evaluation setup, please refer to README.md.