AI & ML
Worst-case bias of the independence assumption for functions of two ordinal variables
arXiv stat.TH0 views
arXiv:2611.24439v1 Announce Type: new Abstract: Let ordinal grades $X$ and $Y$ be obtained by cutting a pair of random variables with copula $C$ at arbitrary thresholds, and let $g$ be a function of the grades. We study the largest error $S_C(g)$, over all thresholds, that is, over all marginal laws, made when the mean of $g(X,Y)$ is computed as if the grades were independent. This is the counterpart of the classical dependence-uncertainty problem: the copula is fixed and the discretisation is free. The error is caused entirely by the interaction of the inputs, measured by an index $\rho(g)$, the largest absolute $g$-volume of a rectangle of grades, which vanishes if and only if $g$ is additive. For every copula, $S_C(g)$ lies between $\rho(g)$ times the largest deviation of $C$ from the independence copula and $\rho(g)$ times half the total variation of their difference, and the factor one half cannot be improved. For the Gaussian copula with correlation $\lambda$ we prove that $S_C(g)$ equals $|\lambda|\rho(g)/(2\pi)$ to first order, with a sharp constant and an explicit remainder. The same first-order law holds for copula families whose first-order departure from independence is a product of unimodal factors, including the Farlie-Gumbel-Morgenstern, Ali-Mikhail-Haq, Frank, Clayton and Plackett families, and we characterise the Archimedean families with this property. An application to the TNM staging of colon cancer, with data from a clinical trial, compares the bounds with the actual error of the independence computation.
Source:https://arxiv.org/abs/2611.24439PreviousParameter estimation for discretely observed rough fractional interacting particle systemsNextAnytime-valid goodness-of-fit test by $e$-spacings
Related posts
A Matching Lower Bound for Online Smooth Calibration
0arXiv stat.ML
How Path Integral Diffusion Decides a Finite Spin Glass
0arXiv stat.ML
Prescriptive Learning of Linear Policies with Feasibility Guarantees for Binary Decision-Making
0arXiv stat.ML
IQS-BO: In-Context Query Selection for Bayesian Optimisation
0arXiv stat.ML
Swiss-Knife: A Framework for Reconfigurable Externalised Multi-Objective Alignment at Decode Time
0arXiv stat.ML
Inverse Design for Conditional Distribution Matching
0arXiv stat.ML