TS-ICL : A Flexible Time-Indexed Foundation Model for Time Series via In-Context Learning#

arXiv PyPI test Python

TS-ICL is a continuous probabilistic Time Series Foundation Model (TSFM) that unifies forecasting and imputation in a single zero-shot architecture, requiring no task-specific training or fine-tuning.


Quickstart

Check out the TS-ICL on-boarding guide.

To the main quickstart

Forecasting

Check out the TS-ICL forecasting guide.

To the forecasting quickstart

Imputation

Check out the TS-ICL imputation guide.

To the imputation quickstart

API reference

The reference guide contains a detailed description of the functions, modules, and objects included in TS-ICL.

To the reference guide


Paper: TS-ICL: A Flexible Time-Indexed Foundation Model for Time Series via In-Context Learning