Aplicação do modelo de Regressão Logística para apoio à decisão de crédito

Abstract

The financial system, from the acceleration of globalization, has come under the influ- ence of many different factors. Thus, specifically, with financial institutions, the need to analyze clients to get a better selection of them, in order to minimize the risks of default has risen. Traditional methods of decision making to provide credit to an individ- ual (or firm) used in particular professional expertises, based on experiences of previous decisions. However, the increased demand for credit resulting from economic pressures, combined with increased commercial competition and the blossoming of new computer technologies, have led to the development of sophisticated statistical techniques capable of distinguishing the applicants for credit as good or bad credit - the model systems of Credit Scoring. Among the many different models cited in the literature and used in practice, the importance of traditional logistic regression model can be acknowledge. The construction of models of Credit Scoring is inserted in the context of data mining, which includes the process of exploration, selection, and modeling large amounts of data to discover regularities or relationships between variables and the logistic regression model is a technique of classification that stands out in this context. This paper will consider the results of Credit Scoring to discriminate indi- viduals into groups (or companies) worthy and not worthy of credit, using the logistic regression model within the context of Data Mining.