A tiered spatial validation workflow for geographic prediction models
Xueqian Jiang, Jialun Wu, Jiabei Liu · 2026
Read A tiered spatial validation workflow for geographic prediction models by Xueqian Jiang, Jialun Wu, Jiabei Liu (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/PMC13470112.txt.
Description
Xueqian Jiang, Jialun Wu, and Jiabei Liu lay out in 2026 a tiered workflow for honestly testing geographic prediction models. Nearby places resemble each other, so random test splits flatter models that will face new regions. Their workflow separates three questions: filling gaps inside a known area, separating clusters within it, and transferring to a distinct region altogether.
Checks stack coordinate clusters, random folds, a size matched random control, network communities, administrative holdouts, and one outside region stress test. A demonstration used 211 Baltimore house sales predicted from building traits. Random scores beat clustered scores, and holdouts of whole districts or regions cut performance further. Cluster based testing sat closer to true outside transfer, though still optimistic. The authors urge matching every accuracy claim to the deployment setting it assumes.
Catalog record
| CREATOR | Xueqian Jiang, Jialun Wu, Jiabei Liu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 23952 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |