Pair-copula based mixture models and their application in clustering

作者:

Highlights:

• A pair-copula based mixture model.

• Separation of margins and dependence structures of a multivariate distribution.

• Design of an EM parameter estimation framework that allows online selection of margins and copulas.

• Application of pair-copula mixture distribution to the problem of model based clustering.

摘要

Highlights•A pair-copula based mixture model.•Separation of margins and dependence structures of a multivariate distribution.•Design of an EM parameter estimation framework that allows online selection of margins and copulas.•Application of pair-copula mixture distribution to the problem of model based clustering.

论文关键词:Pair-copula construction,D-vine,Mixture models,Clustering

论文评审过程:Received 26 November 2012, Revised 26 September 2013, Accepted 2 October 2013, Available online 9 October 2013.

论文官网地址:https://doi.org/10.1016/j.patcog.2013.10.004