# OnlineSTL: Scaling Time Series Decomposition by 100x

arXiv: https://arxiv.org/abs/2107.09110 · HTML: https://mlsanity.com/paper/2107.09110/

## Key facts

- Cited by 5 papers in the map (data through 14 Oct 2026).
- Builds on 3 of its 53 references that are in the map.
- First posted on arXiv 19 Jul 2021.

## Builds on (most cited first, 3 of 3)

- [RobustSTL: A Robust Seasonal-Trend Decomposition Algorithm for Long Time Series](https://mlsanity.com/paper/1812.01767/) (2018)
- [STR: Seasonal-Trend Decomposition Using Regression](https://mlsanity.com/paper/2009.05894/) (2020)
- [ASAP: Prioritizing Attention via Time Series Smoothing](https://mlsanity.com/paper/1703.00983/) (2017)

## Built on by (most cited first, 5 of 5)

- [OneShotSTL: One-Shot Seasonal-Trend Decomposition For Online Time Series Anomaly Detection And Forecasting](https://mlsanity.com/paper/2304.01506/) (2023)
- [Conformal time series decomposition with component-wise exchangeability](https://mlsanity.com/paper/2406.16766/) (2024)
- [Revisiting the Seasonal Trend Decomposition for Enhanced Time Series Forecasting](https://mlsanity.com/paper/2602.18465/) (2026)
- [Ensemble2: Anomaly Detection via EVT-Ensemble Framework for Seasonal KPIs in Communication Network](https://mlsanity.com/paper/2205.14305/) (2022)
- [Frequency-Aware Attention-LSTM for PM2.5 Time Series Forecasting](https://mlsanity.com/paper/2503.24043/) (2025)

## Abstract (arXiv)

Decomposing a complex time series into trend, seasonality, and remainder components is an important primitive that facilitates time series anomaly detection, change point detection, and forecasting. Although numerous batch algorithms are known for time series decomposition, none operate well in an online scalable setting where high throughput and real-time response are paramount. In this paper, we propose OnlineSTL, a novel online algorithm for time series decomposition which is highly scalable and is deployed for real-time metrics monitoring on high-resolution, high-ingest rate data. Experiments on different synthetic and real world time series datasets demonstrate that OnlineSTL achieves orders of magnitude speedups (100x) while maintaining quality of decomposition.