Dirichlet Process Mixtures of Generalized Linear Models

13th International Conference on Artificial Intelligence and Statistics

计算机科学;数学科学

Lauren A. Hannah David M. Blei Warren B. Powell

PID  :  119783

来源: CEUR

【 摘 要 】

We propose Dirichlet Process mixtures of Generalized Linear Models (DPGLMs), a new method of nonparametric regression that accommodates continuous and categorical in puts, models a response variable locally by a generalized linear model. We give conditions for the existence and asymptotic unbiased ness of the DPGLM regression mean func tion estimate; we then give a practical ex ample for when those conditions hold. We evaluate DPGLM on several data sets, com paring it to modern methods of nonparamet ric regression including regression trees and