The Tabular Foundation Model TabPFN Outperforms Specialized Time Series Forecasting Models Based on Simple Features

Shi Bin Hoo, Samuel Müller, David Salinas, Frank Hutter

Keywords: time-series forecasting, foundation model, tabular learning, prior-fitted-networks

TL;DR: We apply TabPFN, a tabular foundation model with simple feature engineering, to time series forecasting and show it outperforms state-of-the-art specialized time series models.

Abstract: Foundation models have become popular in forecasting due to their ability to make accurate predictions, even with minimal fine-tuning on specific datasets. In this paper, we demonstrate how the newly released regression variant of TabPFN, a general tabular foundation model, can be applied to time series forecasting. We propose a straightforward approach, \tabpfnts, which pairs TabPFN with simple feature engineering to achieve strong forecasting performance. Despite its simplicity and with only 11M parameters, \tabpfnts outperforms Chronos-Mini, a model of similar size, and matches or even slightly outperforms Chronos-Large, which has 65-fold more parameters. A key strength of our method lies in its reliance solely on artificial data during pre-training, avoiding the need for large training datasets and eliminating the risk of benchmark contamination.

Submission Number: 35