Θεωρία των Copulas και χρήση τους για παραγωγή συνθετικών δεδομένων
Theory of Copulas and their use to generate synthetic data
View/ Open
Abstract
Over the last few decades, an increasing number of scientists in different scientific fields are getting interested in Copulas. This is mainly due to the fact that in many areas of their research they have to deal with dependent data. In this dissertation, we give some basic definitions to present the theory of Copulas and some of their properties. Furthermore, we present some correlation and dependence indices and we analyse two classes of Copulas, known as Elliptical and Archimedean Copulas. Moreover, we describe algorithms that generate data from specific families, methods to estimate the Copulas and Goodness-of-fit tests for Copulas. Finally, we use simulated data from a specific family of Copulas in order to apply the techniques illustrated in the dissertation.