Taylor & Francis Group
Browse
ubes_a_1721295_sm8046.zip (2.8 MB)

Bayesian Inference for Regression Copulas

Download (2.8 MB)
Version 14 2021-09-29, 16:01
Version 13 2021-09-29, 15:16
Version 12 2021-09-29, 15:15
Version 11 2021-09-29, 15:14
Version 10 2021-09-29, 15:14
Version 9 2021-09-29, 15:13
Version 8 2021-04-01, 11:40
Version 7 2021-01-27, 14:17
Version 6 2020-10-07, 13:30
Version 5 2020-08-24, 08:31
Version 4 2020-05-29, 12:15
Version 3 2020-05-14, 11:20
Version 2 2020-02-26, 13:33
Version 1 2020-01-23, 19:40
dataset
posted on 2021-09-29, 16:01 authored by Michael Stanley Smith, Nadja Klein

We propose a new semiparametric distributional regression smoother that is based on a copula decomposition of the joint distribution of the vector of response values. The copula is high-dimensional and constructed by inversion of a pseudo regression, where the conditional mean and variance are semiparametric functions of covariates modeled using regularized basis functions. By integrating out the basis coefficients, an implicit copula process on the covariate space is obtained, which we call a “regression copula.” We combine this with a nonparametric margin to define a copula model, where the entire distribution—including the mean and variance—of the response is a smooth semiparametric function of the covariates. The copula is estimated using both Hamiltonian Monte Carlo and variational Bayes; the latter of which is scalable to high dimensions. Using real data examples and a simulation study, we illustrate the efficacy of these estimators and the copula model. In a substantive example, we estimate the distribution of half-hourly electricity spot prices as a function of demand and two time covariates using radial bases and horseshoe regularization. The copula model produces distributional estimates that are locally adaptive with respect to the covariates, and predictions that are more accurate than those from benchmark models. Supplementary materials for this article are available online.

Funding

Nadja Klein gratefully acknowledges funding from the Alexander von Humboldt Foundation and the German research foundation (DFG) through the Emmy Noether grant KL 3037/1-1.

History