Multi-Variate Time Series Forecasting on Variable Subsets
This addresses robustness issues in forecasting for applications like sensor networks, but it is incremental as it builds on existing models with a wrapper approach.
The paper tackles the problem of multivariate time series forecasting when only a subset of variables is available during inference, such as due to sensor failures, and shows that state-of-the-art methods degrade significantly in this setting. They propose a non-parametric wrapper technique that recovers close to 95% performance of models even with only 15% of variables present.
We formulate a new inference task in the domain of multivariate time series forecasting (MTSF), called Variable Subset Forecast (VSF), where only a small subset of the variables is available during inference. Variables are absent during inference because of long-term data loss (eg. sensor failures) or high -> low-resource domain shift between train / test. To the best of our knowledge, robustness of MTSF models in presence of such failures, has not been studied in the literature. Through extensive evaluation, we first show that the performance of state of the art methods degrade significantly in the VSF setting. We propose a non-parametric, wrapper technique that can be applied on top any existing forecast models. Through systematic experiments across 4 datasets and 5 forecast models, we show that our technique is able to recover close to 95\% performance of the models even when only 15\% of the original variables are present.