Multiscale detection of practically significant changes in a gradually varying time series
Patrick Bastian, Holger Dette · 2025
Read Multiscale detection of practically significant changes in a gradually varying time series by Patrick Bastian, Holger Dette (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-2504-15872.txt.
Description
Patrick Bastian and Holger Dette build a test for slow drift that grows large enough to matter. Their 2025 paper treats data that changes bit by bit over time, such as sensor or health series. Small drift can be ignored, but drift past a set line calls for action.
Older tests miss such shifts, the authors note, or force users to pick smoothing settings and assume a smooth path. Their method checks many time scales at once and compares later means with an early base stretch. Math proofs back the test, and trials on fake and real data show strong results. The tool fits cases where only a sizable, lasting shift should raise an alarm.
Catalog record
| CREATOR | Patrick Bastian, Holger Dette |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-15872