What question did this study set out to answer?

To enhance interpretability and performance in time-series forecasting by using a model selection framework.

March 8, 2026Machine LearningOpen Access

AALF: Almost Always Linear Forecasting

View Full Paper

Ask AI

Bookmark

Authors

Empirical study shows model selection improves predictive power in time-series forecasting, suggesting simpler models may suffice.

To enhance interpretability and performance in time-series forecasting by using a model selection framework.
Developed a novel online model selection framework for forecasting predictions.
Compared performance of simple models like ARIMA and ETS to Deep Learning models.
Conducted an empirical study across over 3500 individual time-series datasets.
Model selection framework demonstrated competitive performance compared to state-of-the-art methods.
Interpretable models like autoregressive and exponential smoothing performed well, challenging black-box approaches.

Jakobs et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark