Sampling from Dirichlet process mixture models with unknown concentration parameter: mixing issues in large data implementations

• Published in: Statistics and computing Vol. 25; no. 5; pp. 1023 - 1037
• Main Authors: Hastie, David I., Liverani, Silvia, Richardson, Sylvia
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.09.2015
• Subjects: Algorithms; Artificial Intelligence; Computation; Computer simulation; Dirichlet problem; Mathematical models; Mathematics and Statistics; Probability and Statistics in Computer Science; Sampling; Source code; Statistical Theory and Methods; Statistics; Statistics and Computing/Statistics Programs; Mixture model; Profile regression; Dirichlet process; Bayesian clustering
• ISSN: 0960-3174; 1573-1375
• DOI: 10.1007/s11222-014-9471-3

We consider the question of Markov chain Monte Carlo sampling from a general stick-breaking Dirichlet process mixture model, with concentration parameter α . This paper introduces a Gibbs sampling algorithm that combines the slice sampling approach of Walker (Communications in Statistics - Simulation and Computation 36:45–54, 2007 ) and the retrospective sampling approach of Papaspiliopoulos and Roberts (Biometrika 95(1):169–186, 2008 ). Our general algorithm is implemented as efficient open source C++ software, available as an R package, and is based on a blocking strategy similar to that suggested by Papaspiliopoulos (A note on posterior sampling from Dirichlet mixture models, 2008 ) and implemented by Yau et al. (Journal of the Royal Statistical Society, Series B (Statistical Methodology) 73:37–57, 2011 ). We discuss the difficulties of achieving good mixing in MCMC samplers of this nature in large data sets and investigate sensitivity to initialisation. We additionally consider the challenges when an additional layer of hierarchy is added such that joint inference is to be made on α . We introduce a new label-switching move and compute the marginal partition posterior to help to surmount these difficulties. Our work is illustrated using a profile regression (Molitor et al. Biostatistics 11(3):484–498, 2010 ) application, where we demonstrate good mixing behaviour for both synthetic and real examples.