ChronosX: Adapting Pretrained Time Series Models with Exogenous Variables
This addresses the need for more adaptable forecasting models in domains like retail, where covariates such as promotions are critical, though it is incremental in extending pretrained models.
The paper tackles the problem of incorporating exogenous variables into pretrained time series forecasting models, which often lack this capability, and shows that their modular approach outperforms existing baselines on synthetic and real datasets.
Covariates provide valuable information on external factors that influence time series and are critical in many real-world time series forecasting tasks. For example, in retail, covariates may indicate promotions or peak dates such as holiday seasons that heavily influence demand forecasts. Recent advances in pretraining large language model architectures for time series forecasting have led to highly accurate forecasters. However, the majority of these models do not readily use covariates as they are often specific to a certain task or domain. This paper introduces a new method to incorporate covariates into pretrained time series forecasting models. Our proposed approach incorporates covariate information into pretrained forecasting models through modular blocks that inject past and future covariate information, without necessarily modifying the pretrained model in consideration. In order to evaluate our approach, we introduce a benchmark composed of 32 different synthetic datasets with varying dynamics to evaluate the effectivity of forecasting models with covariates. Extensive evaluations on both synthetic and real datasets show that our approach effectively incorporates covariate information into pretrained models, outperforming existing baselines.