Resumen de A model for credit scoring: an application of discriminant analysis

Manuel Artís Ortuño , Montserrat Guillén Estany , José María Martínez

• The application of statistical techniques in decision making, and more specifically for classification requirements, has proved to be adequate in the context of financial problems. In this study, we present the methodology used and the results obtained in the elaboration of a decision-support system for credit assignment. The problem was to provide an automatic tool for a Spanish financial institution that needed to quantify and analyse credit applications from clients. Firstly, we shall present the statistical techniques. Secondly, we will describe the characteristics of the data set used for estimating the discrimination function and, finally, we show the results obtained when the model is used to discriminate among clients in the data set, whose history of financial behaviour is reflected in the data by means of a variable counting the number of unpaid instalments. Essentially, every individual asking the bank for a loan is assigned a certain score. This score is directly related to the probability he or she has of returning the money, but also to the risk of the institution lending that amount. Some comments about the application of the model are given and results concerning the optimal level of risk are also discussed, in order to give clear patterns for implementation