Researchgumbel copulafeature selectionclinical prediction
Gumbel-Copula Selector Improves Clinical Tail Feature Selection
7.1
Researchers propose a supervised feature-selection filter using a Gumbel-copula implied upper-tail concordance score (lambda U) in a paper updated on Feb 24, 2026. They evaluate lambda U on a CDC public-health survey (N=253,680) and the PIMA clinical benchmark (N=768), showing faster selection and reduction from 21 to 10 CDC features with small but significant trade-offs, and comparable or improved ranking versus common baselines.
Key Points
- 1Introduces lambda U, a Gumbel-copula upper-tail concordance score for feature ranking.
- 2Demonstrates better tail-relevant predictor detection than mutual information and mRMR, comparable to ReliefF.
- 3Enables fast, interpretable screening—reduces CDC features 21→10, aiding clinical risk-model preprocessing.
Scoring Rationale
Validated, efficient tail-aware feature selector across large clinical datasets, limited by arXiv preprint status and incremental novelty.
Sources
Public references used for this report.
Practice with real Health & Insurance data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
250 free problems · No credit card
See all Health & Insurance problems