Zero-Shot Time Series Forecasting with Covariates via In-Context Learning
By: Andreas Auer , Raghul Parthipan , Pedro Mercado and more
Potential Business Impact:
Helps predict future events using extra clues.
Pretrained time series models, capable of zero-shot forecasting, have demonstrated significant potential in enhancing both the performance and accessibility of time series forecasting. However, existing pretrained models either do not support covariates or fail to incorporate them effectively. We introduce COSMIC, a zero-shot forecasting model that utilizes covariates via in-context learning. To address the challenge of data scarcity, we propose Informative Covariate Augmentation, which enables the training of COSMIC without requiring any datasets that include covariates. COSMIC achieves state-of-the-art performance in zero-shot forecasting, both with and without covariates. Our quantitative and qualitative analysis demonstrates that COSMIC effectively leverages covariates in zero-shot forecasting.
Similar Papers
ChronosX: Adapting Pretrained Time Series Models with Exogenous Variables
Machine Learning (CS)
Helps predict future sales using extra clues.
Chronos-2: From Univariate to Universal Forecasting
Machine Learning (CS)
Predicts future events using many types of data.
Zero-Shot Forecasting Mortality Rates: A Global Study
Machine Learning (CS)
Predicts future death rates without special training.