Endpoint-preserving correction improves time series forecast accuracy
EPOC: Endpoint-Preserving Online Correction With Compressed Residual State for Multi-Horizon Time Series Forecasting
Machine Learning
Summary
When predicting many steps ahead in time series data, it helps to use past mistakes to improve future forecasts, but keeping all that extra information can be large and slow. The authors propose a new way called EPOC that compresses this extra information and focuses on preserving key points in the data to make corrections. They tested EPOC on multiple datasets and found it reduces prediction errors notably while needing much less memory than comparable methods. This approach balances accuracy improvement and storage efficiency for ongoing time series forecasting.
What this means in practice
- •For financial analysts: improve multi-step market forecasts by applying compact corrections to residual errors without large memory overhead
- •For energy demand planners: enhance long-term electricity load predictions by efficiently updating forecasts based on recent errors with minimal auxiliary data storage
Authors
Takumi Fujimoto, Hiroaki Nishi
Abstract
Completed multi-horizon forecasts provide residual feedback for a fixed forecaster, but retaining full residual blocks increases auxiliary state. We propose Endpoint-Preserving Online Correction (EPOC) with a compressed residual state. It stores low-order discrete cosine transform (DCT) coefficients and the final value of the preceding residual block. Within each channel, the endpoint is shared across component-wise online ridge regressions that also use current-forecast coefficients. The fitted DCT correction is blended with the base forecast. We evaluate eight multivariate series with DLinear and PatchTST, three seeds, and two training variants, yielding 96 matched fixed-base conditions at a 24-step horizon. EPOC achieves mean condition-wise reductions in mean squared error (MSE) and mean absolute error (MAE) of 15.40% and 9.35% from the uncorrected base, respectively, with a median of 6,352 B in retained auxiliary arrays. It has lower paired MSE than the $δ$-Adapter, COSA, FAC, and OMPB in a majority of conditions and uses less state than each. Full ELF achieves the largest mean MSE reduction, 19.29%, but its median retained state is 474,048 B ($\times$75 relative to EPOC). Equal-size summary controls favor the endpoint by 1.65--2.20% in paired MSE; a coefficient-reconstructed endpoint yields similar accuracy to the observed endpoint, highlighting its role as a shared input. Increasing the retained DCT component count from 4 to 8 adds 1.00 percentage point of MSE reduction for 5,728 B. On jointly trained bases, EPOC lowers MSE by 16.69--20.15% relative to globally blended TEFL-style adapters applied to the same base. The code and numerical records are available at https://github.com/keiotakmin/endpoint-preserving-residual-correction.