EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1950. Copula-based synthetic networks generation

Invited abstract in session MA-63: Applications in Finance and Economics, stream OR in Banking, Finance and Insurance: New Tools for Risk Management.

Monday, 8:30-10:00
Room: S14 (building: 101)

Authors (first author is the speaker)

1. Giulia Rotundo
Statistical sciences, Sapienza University of Rome
2. Roy Cerqueti
Department of Social and Economic Sciences, Sapienza University of Rome

Abstract

In this work, we elaborate on the concept of assortative mixing. The role of assortativity contagions has been widely studied: in the spread of epidemics, in systemic risk, and in the resilience of networks, just to point out a few. The classic definition of assortativity through the Pearson correlation coefficient among the degrees of the nodes at the end of each edge captures the linear dependencies among the degrees of the connected nodes. We are going to push forward the concept of assortative mixing allowing for non linear dependencies. In order to achieve this task, we fix the marginal distributions as the most detected from the empirical data sets and we base on the generation of complex networks through copulas. The approach is different from the classic configuration model and its extensions which allow to build a network with a target level of correlation.
It is worth mentioning that the application of the technique goes beyond the mere result of simulating networks.
Actually, the possibility to generate networks with specific properties allows to have a null model for investigating the role of topological structures in specific problems.
For instance, the generation of synthetic data keeping the statistical distribution of real data sets is currently successfully used in machine learning to enlarge the training set.

Keywords

Status: accepted

Back to the list of papers