Teodorescu, Sandra The EM algorithm for a Poisson mixture model. (English) Zbl 1265.65013 An. Univ. București, Mat.-Inform. 56, 133-147 (2007). Summary: We model the distribution of the random variable “number or claims” that affects a policy or a policy portfolio in a determined period of time, using a Poisson mixture. The parameters of the mixture are estimated using two methods: the moments method and the Expectation-Maximization (EM) algorithm used to evaluate maximum likelihood estimators of the model parameters. The methods are illustrate by simulated numerical data. MSC: 65C60 Computational problems in statistics (MSC2010) 62F10 Point estimation Keywords:mixed Poisson distribution; moments method; frequency of damages; policy portfolio; Expectation-Maximization algorithm; maximum likelihood estimators PDFBibTeX XMLCite \textit{S. Teodorescu}, An. Univ. București, Mat.-Inform. 56, 133--147 (2007; Zbl 1265.65013)