Modelarea riscului de credit folosind tehnici de Machine Learning
Credit risk modeling using Machine Learning techniques

Summary/Abstract: Due to the technological advancement associated with Big Data, data availability and computing power, many credit institutions are renewing their business model. Credit risk prediction, monitoring and effective credit processing are key components in the decision-making process of granting the loan. In this paper we have built models of binary classification of the default probability using Machine Learning and Deep Learning procedures, also presenting a ranking of them according to the predictive capacity on out-of-sample data. We noticed that a classification based on Random Forest is superior to Logistic Regression used intensively in banking practice. The paper presents the behavior of these models in a context in which the event of interest has a very low probability of occurrence.

• Details
• Contents