Εκτίμηση ασφαλίστρων με μοντέλα αξιοπιστίας σταυρωτής ταξινόμησης
Premium estimation with cross classification credibility models
Premium estimation is a major issue which insurance companies are faced with. There are lots of methods for premium estimation. One of them is the Credibility technique. In this thesis, firstly, the main concepts of the Credibility Theory and the Analysis of Variance Method are introduced. Afterwards, we present the scientific development of Credibility Models, ending in Hierarchical Model. Simultaneously, all models are approached through the method of Analysis of Variance, in order links between that method and Credibility Theory to be identified. As an inquiry of the Hierarchical Credibility Model, the Cross Classification Model is expounded, in order to cover cases of complex portfolios, while an approach with the use of method of Analysis of Variance is made similarly to the other models. Finally, we present applications of Credibility models and their correlation with the technique of Analysis of Variance, using the statistical package R.