CFE 2019: Start Registration
View Submission - CMStatistics
B1714
Title: A new algorithm for finding a good fitting truncated R-vine copula in high dimensions Authors:  Edith Alice Kovacs - University of Debrecen (Hungary) [presenting]
Tamas Szantai - Budapest University of Technology and Economics (Hungary)
Abstract: Modelling multivariate probability distributions by using R-vine copulas gained popularity due to their flexibility in modelling many types of dependences in the same time. However, in high dimensions, this is also their main drawback, because they involve a large number of parameters. To tackle this problem, two main approaches were proposed in the literature, namely, the truncation and the simplification of the vine copulas. We present a new algorithm for fitting truncated vines. The basic idea is the exploration of the conditional independences defining the truncated vine copula structure. We will show some advantages of the proposed method.