Search papers, labs, and topics across Lattice.
This paper introduces a framework for zero-shot multivariate time series forecasting using tabular foundation models, specifically TabPFN-TS. They reformulate the multivariate time series forecasting problem into a series of scalar regression problems, enabling the application of tabular models without ignoring inter-channel interactions. Experiments demonstrate the effectiveness of this approach compared to state-of-the-art tabular methods.
Tabular foundation models can now tackle multivariate time series forecasting in a zero-shot manner, outperforming existing tabular methods by cleverly reframing the problem as a series of scalar regressions.
Tabular foundation models, particularly Prior-data Fitted Networks like TabPFN have emerged as the leading contender in a myriad of tasks ranging from data imputation to label prediction on the tabular data format surpassing the historical successes of tree-based models. This has led to investigations on their applicability to forecasting time series data which can be formulated as a tabular problem. While recent work to this end has displayed positive results, most works have limited their treatment of multivariate time series problems to several independent univariate time series forecasting subproblems, thus ignoring any inter-channel interactions. Overcoming this limitation, we introduce a generally applicable framework for multivariate time series forecasting using tabular foundation models. We achieve this by recasting the multivariate time series forecasting problem as a series of scalar regression problems which can then be solved zero-shot by any tabular foundation model with regression capabilities. We present results of our method using the TabPFN-TS backbone and compare performance with the current state of the art tabular methods.
