Title: About the estimation of the conditional Kendall's tau and Kendall's Regression
Authors: Alexis Derumigny - University of Twente (Netherlands) [presenting]
Jean-David Fermanian - Ensae-Crest (France)
Abstract: The estimation of the conditional Kendall's tau is considered. The conditional Kendall's tau is a conditional dependence parameter between two variables conditionally to some observed covariates. We propose a nonparametric estimator using kernels. Under a pseudo-GLM specification, we also propose a parametric estimator for the (possibly sparse) vector of coefficients in the model. We study the theoretical properties of both estimators, and prove non-asymptotic bounds that holds with high probability.