University of Ioannina
ΕΚΤΙΜΗΣΗ ΠΑΡΑΜΕΤΡΩΝ ΜΟΝΤΕΛΩΝ ΔΙΑΚΡΙΤΩΝ ΚΑΤΑΝΟΜΩΝ ΜΕ ΤΗ ΜΕΘΟΔΟ ΕΜ
Abstract
dc:descriptionWE CONSIDER THE EM ALGORITHM IN SITUATIONS WHERE ITS DIRECT IMPLEMENTATION REQUIRES AN ITERATIVE PROCESS TO SOLVE THE MAXIMISATION STEP. WE CIRCUMVENT THIS INEFFICIENCY FOR SOME IMPORTANT UNIVARIATE AND BIVARIATE DISTRIBUTIONS BY TAKINGADVANTAGE OF THEIR STRUCTURE AND/OR STATISTICAL PROPERTIES; THE DETAILED PRESENTATION OF THE IDEAS AND METHODS ALLOWS THE READER TO RECOGNISE CASES THAT CAN BE TACKLED SIMILARLY. THUS, FOR EACH CASE CONSIDERED, WE DEVELOP AND POSTULATE A NEW FICTITIOUS COMPLETE-DATA DISTRIBUTION WHICH IS RELATED TO THE ORIGINAL ONE THAT CAUSED THE DIFFICULTY IN THE MAXIMISATION STEP. CONSEQUENTLY, THE INCOMPLETE-DATA PROBLEM IS DEFINED AFRESH AND THE IMPLEMENTATION OF THE EM ALGORITHM IS ACHIEVED BY A RATHER SUCCESSFUL PROXY OF THE ORIGINAL DISTRIBUTION. THE MATERIAL IS SPREAD OVER FOUR CHAPTERS. IN THE FIRST CHAPTER WE DISCUSS THE MAIN POINTS OF THE METHOD OF MAXIMUM LIKELIHOOD ESTIMATION FROM COMPLETE OR INCOMPLETE DATA AND REVIEW THE KEYS RESULTS ASSOCIATED WITH THE EM ALGORITHM. IN THE REMAINING THREE CHAPTERS WE CONSIDER ESTIMATING THE PARAMETERS OF THE POISSON BINOMIAL DISTRIBUTION FROM PARTIALLY GROUPED SAMPLES, THE BIVARIATE POISSON AND BIVARIATE 'SHORT' DISTRIBUTIONS FROM SAMPLES WITH MISSING VALUES AND EXACT OBSERVATIONS RESPECTIVELY. WE INCLUDE ILLUSTRATIONS BASED ON REAL OR SIMULATED SAMPLES AND DISCUSS THE ESTIMATION PROBLEMS IN: THE BIVARIATE POISSON DISTRIBUTION FROM A TRUNCATED SAMPLE, THE NEYMAN TYPE A AND BIVARIATE 'SHORT' II DISTRIBUTIONS FROM GROUPED OBSERVATIONS.
Degree
thesis:*- Grantor dc:publisher
- University of Ioannina
- Year dc:date
- 1994
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Αδαμίδης, Κωνσταντίνος
Subjects
dc:subject × 14Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/3359