Automatic Forecasting via Meta-Learning

Mustafa Abdallah, Ryan Rossi, Kanak Mahadik, Sungchul Kim, Handong Zhao, Haoliang Wang, Saurabh Bagchi

Keywords: Forecasting Model Selection, Time-series Forecasting, Meta-features

Abstract: In this work, we develop techniques for fast automatic selection of the best forecasting model for a new unseen time-series dataset, without having to first train (or evaluate) all the models on the new time-series data to select the best one. In particular, we develop a forecasting meta-learning approach called AutoForecast that allows for the quick inference of the best time-series forecasting model for an unseen dataset. Our approach learns both forecasting models performances over time horizon of same dataset and task similarity across different datasets. The experiments demonstrate the effectiveness of the approach over state-of-the-art (SOTA) single and ensemble methods and several SOTA meta-learners (adapted to our problem) in terms of selecting better forecasting models (i.e., 2X gain) for unseen tasks for univariate and multivariate testbeds.

One-sentence Summary: We develop a method to determine the best model for time-series forecasting on an unseen dataset, without needing exhaustive evaluation of existing models on this dataset.