A Cooperative Classification System for Credit Scoring

Credit scoring (CS) is an important process in both banking and finance. CS is a score measure that helps lenders and creditors to see how good the borrower is. It indicates how likely the borrower is to pay back the owed debt, based on its past borrowing behavior. For instance, this score can be used to avoid risk and provide an objective analysis of the applicant's creditworthiness which reduces credit risk. Also CS allows the automation of the lending process which leads to increase the speed and consistency of the borrower's application. In this paper, we propose to use agents' technology to develop a cooperative classification system for credit scoring. Empirical studies are conducted on five well-known financial datasets. We distribute the complex classification problem on a set of intelligent and cooperative agents where each agent evaluates a given machine learning algorithm on the considered dataset. The use of agents in the classification process provides a high degree of flexibility and allows for a faster time scale of data analysis.