Research map / Time series / Causal and generative time series
RobustSTL: A Robust Seasonal-Trend Decomposition Algorithm for Long Time Series
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use RobustSTL instead of TBATS | Synthetic data trend decomposition · Synthetic data — trend | MSE: 0.3150 -> 0.0530 | Helped |
| Use RobustSTL instead of TBATS | Synthetic data trend decomposition · Synthetic data — trend | MAE: 0.3637 -> 0.1627 | Helped |
| Use RobustSTL instead of TBATS | Synthetic data season decomposition · Synthetic data — season | MSE: 0.1839 -> 0.0265 | Helped |
| Use RobustSTL instead of TBATS | Synthetic data season decomposition · Synthetic data — season | MAE: 0.2770 -> 0.0750 | Helped |
| Combine LAD trend extraction with sparse regularization and non-local seasonal filtering | RobustSTL · Synthetic data with trend shifts, seasonal shifts, anomalies, and noise | Trend and season MSE: RobustSTL: trend 0.0530, season 0.0265; TBATS: trend 0.3150, season 0.1839 | Helped |
| Use RobustSTL rather than standard STL, TBATS, or STR | Real-world data decomposition · Real dataset 1 and real dataset 2 | Trend and season component fidelity | Helped |
| Use bilateral filtering rather than ordinary smoothing that destroys abrupt structure | RobustSTL · Synthetic and real-world time series | Preservation of abrupt trend changes and spikes | Helped |
| Use neighborhoods around prior seasonal cycles instead of only same-phase historical points | RobustSTL non-local seasonal filtering · Time series with seasonality shifts | Seasonality shift handling | Helped |
Closest papers by meaning
- RobustTSF: Towards Theory and Design of Robust Time Series Forecasting with Anomalies (ICLR 2024 · similarity 0.87)
- Fortifying Time Series: DTW-Certified Robust Anomaly Detection (NeurIPS 2025 · similarity 0.86)
- TimeEmb: A Lightweight Static-Dynamic Disentanglement Framework for Time Series Forecasting (NeurIPS 2025 · similarity 0.85)
- Revitalizing Multivariate Time Series Forecasting: Learnable Decomposition with Inter-Series Dependencies and Intra-Series Variations Modeling (ICML 2024 · similarity 0.85)
- Retrieval Over Training: Similarity search-based Model Selection for Time Series Anomaly Detection (NeurIPS 2026 · similarity 0.84)
- Parsimony or Capability? Decomposition Delivers Both in Long-term Time Series Forecasting (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.