Time-series forecasting accuracy depends on synthetic data assembly methods
Evaluating Sequence Assembly Strategies for Differentially Private Synthetic Time-Series Forecasting
Machine Learning
Summary
When creating fake time-series data to protect privacy, the way small pieces of data are joined matters a lot for forecasting models. The authors found that simply improving how these pieces fit together or how similar each piece is to real data does not always make forecasts better. Different forecasting models prefer different ways of stitching these data pieces. Their work helps show how to best assemble synthetic data for better prediction results without changing the data generator itself.
What this means in practice
- •For data privacy engineers: Optimize how synthetic time-series data windows are combined to improve forecasting model training under privacy constraints.
- •For energy demand planners: Use better assembly methods for synthetic utility consumption data to enhance forecasting accuracy while preserving user privacy.
Authors
Guoxiong Long, Huizhen Huang, Qikun Cai, Tao Huang, Chen Hou
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.