The traditional methods of making credit decisions relied mostly on human judgment; those have been replaced by methods that use statistical models. Today, statistical models are used not only for deciding whether to accept an applicant (application scoring), but also to predict the likelihood of defaults among customers who have already been accepted (behavioral scoring) and to predict the likely amount of debt that the lender can expect to recover (collection scoring).
In this paper, Credit Scoring for SAS Enterprise Miner software is used to build credit scoring models for the retail credit industry. In the introduction, it discusses the benefits of performing credit scoring and the advantages of building credit scoring models in-house using SAS Enterprise Miner. The paper also discusses the advantages and disadvantages of three important model types: the scorecard, the decision tree and the neural network. Finally, it includes a case study where an application scoring model is built with SAS Enterprise Miner, beginning with reading the development sample, through classing and selecting characteristics, fitting a regression model, calculating score points, assessing scorecard quality (in comparison to a decision tree model built on the same sample) and going through a reject inference process to arrive at a model for scoring the new customer applicant population.
Efficiency is gained by completely automating the modeling process, but even more so by providing the analyst with a graphical user interface that structures, connects and documents the flow of activities that are carried out. If changes or variations need to be introduced, the overall process is already defined and doesn’t have to be started from scratch. (For example, if a similar analysis needs to be carried out on different data, for a different purpose and by a new analyst, the process is still easy to apply.) Process flows enable the organization to implement its traditional way of working but also to experiment with new approaches and compare the results. The environment is flexible and open to allow the analyst to interact with the data and models and bring in his or her domain expertise.
Review the process and the case study in this best practices paper: Building Credit Scorecards Using Credit Scoring for SAS® Enterprise MinerTM. Do you have other suggestions for papers on credit scoring for your fellow SAS users?
