Decision tree applied in classifying the occurrence of cyber claims in banking sector companies

Abstract

The study aimed to predict cyber claims in companies in the banking sector using a decision tree. To this end, 683 cases of cyber losses were extracted from an operational risk database. The independent variables considered in the modeling were the region of domicile, the size of the company and, as main explanatory variable, revenue. The classification reached 89% of global hits. The modeling in question guarantees a good classification quality and better fit when compared to traditional GLM modeling. The results of this work are useful and can act in an innovative way as a tool to support the decision making of insurers, aiming at useful responses to the management of cyber risks.