CMStatistics 2016: Start Registration
View Submission - CMStatistics
B1387
Title: Two layer EM algorithm for ALD mixture regression models: A new solution to composite quantile regression Authors:  Shanshan Wang - Nanyang Technological University (Singapore)
Liming Xiang - Nanyang Technological University (Singapore) [presenting]
Abstract: Motivated by the link between quantile regression and a likelihood-based approach under assumption of asymmetric Laplace distributed errors, we introduce linear regression by modeling the error term through a nite mixture of asymmetric Laplace distributions. The model expands the flexibility of the linear regression by accounting for heterogeneity among data, and allows us to establish the equivalence between maximum likelihood estimation of the model parameters and the composite quantile regression (CQR) estimation, providing a new likelihood-based solution to CQR. We propose a two-layer EM-based algorithm for implementing the estimation procedure. An appealing feature of the proposed algorithm is that the closed form updates for the parameters in each iteration are obtained explicitly, instead of resorting to linear programming optimization methods, as in the existing work. The computational complexity can be reduced significantly. We evaluate the performance through simulation studies and illustrate its usefulness by analyzing a gene expression dataset.