Open Models Almanac
Methodsdata curationdata filtering

single mention · data curation

PCA-based dimension selection for decorrelated quality signals

A PCA-informed subset of quality dimensions is selected to reduce signal correlation before forming a composite score.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See data filtering for the methods filed beside it.

Also called PCA-informed subset of its dimensions.

Filed under data curation › data filtering