Estimating Mixture of Dirichlet Process Models

MacEachern, Steven N.; Müller, Peter

doi:10.1080/10618600.1998.10474772

Cited by 369 publications

(275 citation statements)

References 16 publications

Supporting

Mentioning

275

Contrasting

Order By: Relevance

“…With a finite number of clusters, this corresponds to fitting a finite mixture of distributions to the data (McLachlan and Peel, 2000), through likelihood-based methods such as expectation-maximization (Dempster et al, 1977) or Markov chain Monte Carlo (MCMC) methods in the Bayesian framework (Diebolt and Robert, 1994). The model can be generalized to a countably infinite number of clusters, by using a prior such as the Dirichlet process prior for mixture components that allows a few components to dominate (MacEachern and Müller, 1998). Such methods have become feasible with the development of powerful MCMC tools.…”

Section: Clustering Methodsmentioning

confidence: 99%

An evolutionary Monte Carlo algorithm for Bayesian block clustering of data matrices

Gupta

2014

Computational Statistics & Data Analysis

View full text Add to dashboard Cite

Section: Clustering Methodsmentioning

confidence: 99%

An evolutionary Monte Carlo algorithm for Bayesian block clustering of data matrices

Gupta

2014

Computational Statistics & Data Analysis

View full text Add to dashboard Cite

“…1 possible values and the other parameters can be updated using standard techniques for hierarchical models. Alternative efficient computational methods have been proposed for non-conjugate models (MacEachern and Müller 1998;Neal 2000).…”

Section: Hðbþð1àhðbþþ Mþ1mentioning

confidence: 99%

Bayesian clustering of distributions in stochastic frontier analysis

Griffin

2011

J Prod Anal

View full text Add to dashboard Cite

In stochastic frontier analysis, firm-specific efficiencies and their distribution are often main variables of interest. If firms fall into several groups, it is natural to allow each group to have its own distribution. This paper considers a method for nonparametrically modelling these distributions using Dirichlet processes. A common problem when applying nonparametric methods to grouped data is small sample sizes for some groups which can lead to poor inference. Methods that allow dependence between each group's distribution are one set of solutions. The proposed model clusters the groups and assumes that the unknown distribution for each group in a cluster are the same. These clusters are inferred from the data. Markov chain Monte Carlo methods are necessary for model-fitting and efficient methods are described. The model is illustrated on a cost frontier application to US hospitals.

show abstract

“…Escobar and West (1995) provided a Gibbs sampling algorithm for the estimation of posterior distribution for all model parameters, MacEachern and Müller (1998) presented a Gibbs sampler with non-conjugate priors by using auxiliary parameters, and Neal (2000) provided an extended and more efficient Gibbs sampler to handle general Dirichlet process mixture models. Teh et al (2006) also extended the auxiliary variable method of Escobar and West (1995) for posterior sampling of the precision parameter with a gamma prior.…”

Section: Sampling Schemes For Glmdmsmentioning

confidence: 99%