Comparing two populations using Bayesian Fourier series density estimation
Marco Henrique de Almeida Inácio, Rafael Izbicki, Luis Ernesto Bueno Salasar · Communications in Statistics - Simulation and Computation · 2018
Marco Henrique de Almeida Inácioab*, Rafael Izbickia & Luis Ernesto Salasaraa Department of Statistics, Federal University of São Carlos (UFSCar), São Carlos, Brazil; b Institute of Mathematics and Computer Sciences, University of São Paulo (USP), São Paulo, BrazilColor versions of one or more of the figures in the article can be found online at www.tandfonline.com/lssp.Supplemental data for this article can be accessed on the publisher's websiteCONTACT Marco Henrique de Almeida Inácio [email protected] Department of Statistics, Federal University of São Carlos (UFSCar), Rodovia Washington Lus, km 235 - SP-310 - São Carlos, Brazil.AbstractAn important question in sciences is how to evaluate the similarity between two populations given independent samples from each of them. The most common approach to solve this is to use standard hypotheses tests. We propose an alternative method to compare two groups using a Bayesian nonparametric framework. The key idea is to measure the similarity between them by evaluating the distance between their associated densities. The nonparametric perspective makes it straightforward to assess the uncertainty about such distance without making strong assumptions about the generating processes. We provide both simulated and real examples to illustrate our method effectiveness.