.png){kind=icon}
.png){kind=icon}
Return
Statistical agnostic regression: A machine learning method to validate regression models
.png){kind=icon}
OA
AI
DOI：10.1016/j.jare.2025.04.026.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• A novel approach, SAR evaluates statistical significance in ML-based linear regression models by analyzing concentration inequalities of the expected loss (actual risk). • SAR introduces a threshold ensuring evidence of a linear relationship in the population, with a probability of at least 1-η1 - η1-η, under non-parametric assumptions. • Simulations show SAR can emulate the classical multivariate FFF-test for slope parameters, offering comparable analyses of variance without relying on traditional assumptions. • Residuals computed from SAR balance characteristics of ML-based and classical OLS residuals, bridging gaps between these methodologies.
Keywords:
Ordinary least squares
K-fold cross-validation
linear support vector machines
Statistical learning theory
Permutation tests
Upper bounding
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
13
Papers:
2.8K
Citations:
1.4W