# The Stationarity Bias: Stratified Stress-Testing for Time-Series Imputation in Regulated Dynamical Systems

arXiv: https://arxiv.org/abs/2602.15637 · HTML: https://mlsanity.com/paper/2602.15637/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 10 of its 32 references that are in the map.
- First posted on arXiv 17 Feb 2026.

## Builds on (most cited first, 10 of 10)

- [Optuna: A Next-generation Hyperparameter Optimization Framework](https://mlsanity.com/paper/1907.10902/) (2019)
- [One Fits All:Power General Time Series Analysis by Pretrained LM](https://mlsanity.com/paper/2302.11939/) (2023)
- [GAIN: Missing Data Imputation using Generative Adversarial Nets](https://mlsanity.com/paper/1806.02920/) (2018)
- [SCINet: Time Series Modeling and Forecasting with Sample Convolution and Interaction](https://mlsanity.com/paper/2106.09305/) (2021)
- [GP-VAE: Deep Probabilistic Time Series Imputation](https://mlsanity.com/paper/1907.04155/) (2019)
- [SAITS: Self-Attention-based Imputation for Time Series](https://mlsanity.com/paper/2202.08516/) (2022)
- [TSLANet: Rethinking Transformers for Time Series Representation Learning](https://mlsanity.com/paper/2404.08472/) (2024)
- [TSI-Bench: Benchmarking Time Series Imputation](https://mlsanity.com/paper/2406.12747/) (2024)
- [Time Evidence Fusion Network: Multi-source View in Long-Term Time Series Forecasting](https://mlsanity.com/paper/2405.06419/) (2024)
- [The Driver-Blindness Phenomenon: Why Deep Sequence Models Default to Autocorrelation in Blood Glucose Forecasting](https://mlsanity.com/paper/2511.20601/) (2025)

## Built on by (most cited first, 1 of 1)

- [Interventional Flow Matching: Prospective Dose-Response Forecasting with Velocity-Field Jacobian Regularization](https://mlsanity.com/paper/2606.29386/) (2026)

## Abstract (arXiv)

Time-series imputation benchmarks employ uniform random masking and shape-agnostic metrics (MSE, RMSE), implicitly weighting evaluation by regime prevalence. In systems with a dominant attractor -- homeostatic physiology, nominal industrial operation, stable network traffic -- this creates a systematic Stationarity Bias: simple methods appear superior because the benchmark predominantly samples the easy, low-entropy regime where they trivially succeed. We formalize this bias and propose a Stratified Stress-Test that partitions evaluation into Stationary and Transient regimes. Using Continuous Glucose Monitoring (CGM) as a testbed -- chosen for its rigorous ground-truth forcing functions (meals, insulin) that enable precise regime identification -- we establish three findings with broad implications:(i)~Stationary Efficiency: Linear interpolation achieves state-of-the-art reconstruction during stable intervals, confirming that complex architectures are computationally wasteful in low-entropy regimes.(ii)~Transient Fidelity: During critical transients (post-prandial peaks, hypoglycemic events), linear methods exhibit drastically degraded morphological fidelity (DTW), disproportionate to their RMSE -- a phenomenon we term the RMSE Mirage, where low pointwise error masks the destruction of signal shape.(iii)~Regime-Conditional Model Selection: Deep learning models preserve both pointwise accuracy and morphological integrity during transients, making them essential for safety-critical downstream tasks. We further derive empirical missingness distributions from clinical trials and impose them on complete training data, preventing models from exploiting unrealistically clean observations and encouraging robustness under real-world missingness. This framework generalizes to any regulated system where routine stationarity dominates critical transients.