Historical Inertia: A Neglected but Powerful Baseline for Long Sequence Time-series Forecasting

TL;DR

This paper introduces historical inertia as a simple yet powerful baseline for long sequence time-series forecasting, showing it can significantly outperform more complex models.

Contribution

The paper proposes a new baseline called historical inertia for LSTF, emphasizing the importance of recent historical data in forecasting accuracy.

Findings

Up to 82% relative improvement over state-of-the-art methods.

Historical inertia alone can serve as a strong baseline.

Experimental validation on four real-world datasets.

Abstract

Long sequence time-series forecasting (LSTF) has become increasingly popular for its wide range of applications. Though superior models have been proposed to enhance the prediction effectiveness and efficiency, it is reckless to neglect or underestimate one of the most natural and basic temporal properties of time-series. In this paper, we introduce a new baseline for LSTF, the historical inertia (HI), which refers to the most recent historical data-points in the input time series. We experimentally evaluate the power of historical inertia on four public real-word datasets. The results demonstrate that up to 82\% relative improvement over state-of-the-art works can be achieved even by adopting HI directly as output.