A Classification Knowledge Acquisition of Integrated Rough Sets Classifiers in the Banking Industry

Today, the global issuers and investors of financial systems face key economic challenges of financial crisis. It will turn into banking panics and jeopardize the economic stability much more likely that when banks become insolvent or bankrupt, especially in large banks. In order to offer a suitable assess of the financial and operational competence of banks to interested parties, credit rating agencies provide an opinion for a capacity of an entity that performs financial commitments. Furthermore, many models had been proposed to deal with credit rating classification problems, but these models needed to satisfy specific assumptions about data distributions or lack of explanation power of decision rules for the results. Therefore, the study aims to propose a MEPA (minimize entropy principle approach)-based rough sets classifier procedure, which uses MEPA method and rough set theory to improve classification accuracy and extract comprehensive decision rules for interested parties. For verification, a dataset of 1,950 samples collected from BANKSCOPE database is used, which included 420 large banks in the global financial markets during 1998-2007. The experimental results reveal that the proposed procedure outperforms the listed models in terms of accuracy.