OLinear: A Linear Model for Time Series Forecasting in Orthogonally Transformed Domain
Wenzhen Yue, Yong Liu, Haoxuan Li, Hao Wang, Xianghua Ying, Ruohao Guo, Bowei Xing, Ji Shi · 2025
Read OLinear: A Linear Model for Time Series Forecasting in Orthogonally Transformed Domain by Wenzhen Yue, Yong Liu, Haoxuan Li, Hao Wang, Xianghua Ying, Ruohao Guo, Bowei Xing, Ji Shi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-08550.txt.
Introduction
Wenzhen Yue and co-authors present in 2025 a plain linear forecaster for time series that works after a smart change of basis. Most forecasters read and write the series in raw time order. There, tangled step to step links can trip even deep models.
Some rivals switch to fixed bases like sine waves, which ignore the data at hand. This team's transform is learned from the data: an orthogonal matrix that lines up with the series' own variance pattern. A linear model then forecasts in that cleaned domain. Simple models that beat deep ones on such tasks keep embarrassing the field. A data fitted basis may be why this one holds up across sets.
Catalog record
| CREATOR | Wenzhen Yue, Yong Liu, Haoxuan Li, Hao Wang, Xianghua Ying, Ruohao Guo, Bowei Xing, Ji Shi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 186 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-08550