Reviving Error Correction in Modern Deep Time-Series Forecasting

TL;DR

This paper introduces UEC-STD, a universal, architecture-agnostic error correction method for deep time-series forecasting that improves long-term prediction accuracy by explicitly correcting trend and seasonal components.

Contribution

The authors propose a novel, simple error correction model that can be integrated with any deep forecaster without retraining, significantly enhancing long-term forecast accuracy.

Findings

UEC-STD improves correction accuracy across 4 backbones and 10 datasets.

Explicit decomposition into trend and seasonal components enhances correction robustness.

The method is architecture-agnostic and easily integrable without retraining.

Abstract

Modern deep-learning models have achieved remarkable success in time-series forecasting. Yet, their performance degrades in long-term prediction due to error accumulation in autoregressive inference, where predictions are recursively used as inputs. While classical error correction mechanisms (ECMs) have long been used in statistical methods, their applicability to deep learning models remains limited or ineffective. In this work, we revisit the error accumulation problem in deep time-series forecasting and investigate the role and necessity of ECMs in this new context. We propose a simple, architecture-agnostic error correction model that can be integrated with any existing forecaster without requiring retraining. By explicitly decomposing predictions into trend and seasonal components and training the corrector to adjust each separately, we introduce the Universal Error Corrector with Seasonal-Trend Decomposition (UEC-STD), which significantly improves correction accuracy and robustness across 4 backbones and 10 datasets. Our findings provide a practical tool for enhancing forecasts while offering new insights into mitigating autoregressive errors in deep time-series models. Code is available at https://github.com/DA2I2-SLM/UEC-STD.