Evaluating Sequence Assembly Strategies for Differentially Private Synthetic Time-Series Forecasting

2026-10-07

AI Summary

This study addresses the unclear impact of fixed-window assembly strategies on downstream forecasting utility in differentially private synthetic time series. We systematically evaluate how overlap ratios and weighting schemes affect boundary continuity and statistical fidelity, conducting multi-model comparative experiments within the Train on Synthetic, Test on Real (TSTR) framework to quantify the dependency between assembly configurations and forecasting performance. Our findings reveal that the optimal assembly strategy is fundamentally contingent upon the specific forecaster employed, demonstrating that no single diagnostic metric suffices to guide configuration selection. Furthermore, we validate the robustness of this principle and establish forecaster-specific optimal assembly paradigms, providing critical practical guidance for privacy-preserving time series generation.

Abstract

Differentially private time-series generators commonly produce fixed-length synthetic windows, whereas downstream forecasting models often require long continuous training sequences. How these windows are assembled after generation can therefore alter the effective synthetic data presented to a forecaster, even when the trained generator remains unchanged. We study this post-generation sequence assembly process by systematically varying overlap rates and window-weighting schemes and evaluating the resulting sequences in terms of boundary continuity, statistical and temporal fidelity, and Train-on-Synthetic-Test-on-Real (TSTR) forecasting utility. Across four types of public datasets (ETTh1, ETTm1, Weather, and Appliances) and five forecasting models, the results reveal a clear forecaster-dependent assembly principle: downstream TSTR utility is jointly shaped by the forecaster, overlap rate, and window-weighting scheme, leading to distinct assembly preferences across forecasting models. Increased overlap generally improves boundary continuity, but improvements in continuity or individual fidelity diagnostics do not consistently reduce forecasting error, indicating that these diagnostics alone are insufficient for selecting assembly configurations. Complete five-forecaster assembly grids, together with matched Train-on-Real-Test-on-Real (TRTR) references, further characterize these regularities and quantify assembly-dependent utility relative to real-data training. We then validate the identified principles through additional analyses of robustness and generator variability.

Problem

• differential privacy
• synthetic time-series
• sequence assembly
• forecasting utility
• TSTR

Innovation

• Differential Privacy
• Synthetic Time-Series
• Sequence Assembly
• TSTR Utility
• Time-Series Forecasting