From Similarity to Superiority: Channel Clustering for Time Series Forecasting

TL;DR

This paper introduces a novel Channel Clustering Module (CCM) that dynamically groups similar channels in time series data, enhancing forecasting accuracy, enabling zero-shot predictions, and improving model interpretability by balancing channel independence and dependence strategies.

Contribution

The paper proposes a new CCM that effectively balances channel independence and dependence by clustering similar channels, improving forecasting performance and interpretability.

Findings

CCM boosts forecasting accuracy by 2.4% (long-term) and 7.2% (short-term) on average.

Enables zero-shot forecasting with existing models.

Reveals intrinsic channel similarities, enhancing interpretability.

Abstract

Time series forecasting has attracted significant attention in recent decades. Previous studies have demonstrated that the Channel-Independent (CI) strategy improves forecasting performance by treating different channels individually, while it leads to poor generalization on unseen instances and ignores potentially necessary interactions between channels. Conversely, the Channel-Dependent (CD) strategy mixes all channels with even irrelevant and indiscriminate information, which, however, results in oversmoothing issues and limits forecasting accuracy. There is a lack of channel strategy that effectively balances individual channel treatment for improved forecasting performance without overlooking essential interactions between channels. Motivated by our observation of a correlation between the time series model's performance boost against channel mixing and the intrinsic similarity on a pair of channels, we developed a novel and adaptable Channel Clustering Module (CCM). CCM dynamically groups channels characterized by intrinsic similarities and leverages cluster information instead of individual channel identities, combining the best of CD and CI worlds. Extensive experiments on real-world datasets demonstrate that CCM can (1) boost the performance of CI and CD models by an average margin of 2.4% and 7.2% on long-term and short-term forecasting, respectively; (2) enable zero-shot forecasting with mainstream time series forecasting models; (3) uncover intrinsic time series patterns among channels and improve interpretability of complex time series models.