A likelihood-based constrained algorithm for multivariate normal mixture models. (English) Zbl 1205.62066

Ingrassian, Salvatore

Stat. Methods Appl. 13, No. 2, 151-166 (2004).

Summary: It is well known that the log-likelihood function for samples coming from normal mixture distributions may present spurious maxima and singularities. For this reason we reformulate some Hathaway’s results and we propose two constrained estimation procedures for multivariate normal mixture modelling according to the likelihood approach. Their performances are illustrated on the grounds of some numerical simulations based on the EM algorithm. A comparison between multivariate normal mixtures and the hot-deck approach in missing data imputation is also considered.

MSC:

62H12	Estimation in multivariate analysis
65C60	Computational problems in statistics (MSC2010)

Keywords:

EM algorithm; missing data