Refactor Analysis: Powerful and Predictive Evaluations of Factor Models and Dimensionality Journal: Data/Math Publication type: STATISTICAL METHODS [preprint]
Michael Hardy · 2026
Read Refactor Analysis: Powerful and Predictive Evaluations of Factor Models and Dimensionality Journal: Data/Math Publication type: STATISTICAL METHODS [preprint] by Michael Hardy (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/arxiv-2603-20938.txt.
Introduction
Hardy introduces Refactor analysis, a data first way to test one factor models. Single scores and leaderboards rest on such models, yet fit is usually judged on correlation matrices, not the raw responses. Refactor turns a one factor solution into a direct prediction of the original data matrix. A cross validated variant, Verifactor, checks how well it generalizes. On two hundred public binary data sets, classic fit measures relate weakly to how well the data itself is recovered. Quadrant correlation also beats common choices for reconstruction.
Catalog record
| CREATOR | Michael Hardy |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 166743 |
| LICENSE | http://creativecommons.org/licenses/by-sa/4.0/ |
| RIGHTS | CC BY SA |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.20938