Feature-weighted maximum representative subsampling
Tony Hauptmann, Stefan Kramer · 2026
Read Feature-weighted maximum representative subsampling by Tony Hauptmann, Stefan Kramer (2026) 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/pmc-13234176.txt.
Description
Tony Hauptmann and Stefan Kramer published a 2026 statistics study on fixing biased survey samples without wrecking the good parts. Social science datasets often need reweighting so they match a trusted reference sample. When only a few variables are badly skewed, standard debiasing distorts everything else. The authors' method, feature-weighted maximum representative subsampling, trims a sample toward representativeness while deliberately going easy on the worst-biased features.
Weights for each feature come from a classifier trained to tell the two datasets apart, so important differences still count. On eight artificially biased datasets, the method kept more usable records, and downstream prediction quality showed no significant loss. A real social science dataset confirmed the approach. The code is public, and the authors frame the trade-off as keeping data that heavy-handed correction would throw away.
Catalog record
| CREATOR | Tony Hauptmann, Stefan Kramer |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 48 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |