Construction of Probabilistic Boolean Network for Credit Default Data

Abstract

In this article, we consider the problem of construction of Probabilistic Boolean Networks (PBNs). Previous works have shown that Boolean Networks (BNs) and PBNs have many potential applications in modeling genetic regulatory networks and credit default data. A PBN can be considered as a Markov chain process and the construction of a PBN is an inverse problem. Given the transition probability matrix of the PBN, we try to find a set of BNs with probabilities constituting the given PBN. We propose a revised estimation method based on entropy approach to estimate the model parameters. Practical real credit default data are employed to demonstrate our proposed method.