Zero-shot Multivariate Time Series Forecasting Using Tabular Prior Fitted Networks
2026-04-09 • Machine Learning
Machine LearningArtificial Intelligence
AI summaryⓘ
The authors looked at special AI models called tabular foundation models, which work well with table-like data, and explored how to use them for forecasting multiple time series together. Usually, people treat multiple time series as separate single series and miss how they relate to each other. The authors created a new way to turn the multivariate forecasting problem into many simple prediction tasks that these models can handle without extra training. They tested their method with a specific model called TabPFN-TS and compared it to other top tabular methods.
tabular datafoundation modelstime series forecastingmultivariate time seriesscalar regressionTabPFNzero-shot learningdata imputationregression tasks
Authors
Mayuka Jayawardhana, Nihal Sharma, Kazem Meidani, Bayan Bruss, Tom Goldstein, Doron Bergman
Abstract
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.