New framework estimates ROC surface volumes and inequality

This paper introduces a novel framework for calculating the volumes of sets that are projections of critical function sets, with a specific focus on the convex body underlying the optimal receiver operating characteristic (ROC) surface. The proposed method utilizes an Aumann expectation representation and Minkowski mixed volumes to determine the volume under the ROC surface (VUS). Researchers developed a double/debiased machine learning estimator for VUS, detailing its asymptotic properties and an inference procedure. The framework also extends to analyzing feasible error sets across different groups and offers a generalized Gini coefficient for inequality measurement. AI

IMPACT Introduces novel statistical methods applicable to machine learning, potentially improving model evaluation and inequality analysis.

RANK_REASON The item is an academic paper published on arXiv detailing a new statistical framework and estimation method. [lever_c_demoted from research: ic=1 ai=0.7]

Read on arXiv stat.ML →

• arXiv
• Aumann expectation
• CatalyzeX Code Finder for Papers
• CORE Recommender
• DagsHub
• Gini coefficient
• Gotit.pub
• Hugging Face
• Influence Flower
• Minkowski mixed volumes
• ScienceCast
• U-statistic kernel

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →