Wasserstein Regression with Empirical Measures and Density Estimation for Sparse Data

TL;DR

This paper introduces a novel Wasserstein regression method using empirical measures and density estimation, effectively handling sparse data situations where traditional density estimation fails, and demonstrates superior performance through simulations and real-world case studies.

Contribution

It proposes a new distribution-response regression approach based on empirical measures that works well with sparse data and avoids complex density estimation issues.

Findings

Outperforms existing methods in simulations.

Effective with small sample sizes for distributions.

Validated on environmental and auction datasets.

Abstract

The problem of modeling the relationship between univariate distributions and one or more explanatory variables has found increasing interest. Traditional functional data methods cannot be applied directly to distributional data because of their inherent constraints. Modeling distributions as elements of the Wasserstein space, a geodesic metric space equipped with the Wasserstein metric that is related to optimal transport, is attractive for statistical applications. Existing approaches proceed by substituting proxy estimated distributions for the typically unknown response distributions. These estimates are obtained from available data but are problematic when for some of the distributions only few data are available. Such situations are common in practice and cannot be addressed with available approaches, especially when one aims at density estimates. We show how this and other problems associated with density estimation such as tuning parameter selection and bias issues can be side-stepped when covariates are available. We also introduce a novel version of distribution-response regression that is based on empirical measures. By avoiding the preprocessing step of recovering complete individual response distributions, the proposed approach is applicable when the sample size available for some of the distributions is small. In this case, one can still obtain consistent distribution estimates even for distributions with only few data by gaining strength across the entire sample of distributions, while traditional approaches where distributions or densities are estimated individually fail, since sparsely sampled densities cannot be consistently estimated. The proposed model is demonstrated to outperform existing approaches through simulations. Its efficacy is corroborated in two case studies on Environmental Influences on Child Health Outcomes (ECHO) data and eBay auction data.