From Developing Credit Risk Models Using SAS Enterprise Miner and SAS/STAT. Full book available for purchase here.

Transcription

1 From Developing Credit Risk Models Using SAS Enterprise Miner and SAS/STAT. Full book available for purchase here. About this Book... ix About the Author... xiii Acknowledgments...xv Chapter 1 Introduction Book Overview Overview of Credit Risk Modeling Regulatory Environment Minimum Capital Requirements Expected Loss Unexpected Loss Risk Weighted Assets SAS Software Utilized Chapter Summary References and Further Reading Chapter 2 Sampling and Data Pre-Processing Introduction Sampling and Variable Selection Sampling Variable Selection Missing Values and Outlier Treatment Missing Values Outlier Detection Data Segmentation Decision Trees for Segmentation K-Means Clustering... 24

2 iv 2.5 Chapter Summary References and Further Reading Chapter 3 Development of a Probability of Default (PD) Model Overview of Probability of Default PD Models for Retail Credit PD Models for Corporate Credit PD Calibration Classification Techniques for PD Logistic Regression Linear and Quadratic Discriminant Analysis Neural Networks Decision Trees Memory Based Reasoning Random Forests Gradient Boosting Model Development (Application Scorecards) Motivation for Application Scorecards Developing a PD Model for Application Scoring Model Development (Behavioral Scoring) Motivation for Behavioral Scorecards Developing a PD Model for Behavioral Scoring PD Model Reporting Overview Variable Worth Statistics Scorecard Strength Model Performance Measures Tuning the Model Model Deployment Creating a Model Package Registering a Model Package Chapter Summary References and Further Reading... 58

3 v Chapter 4 Development of a Loss Given Default (LGD) Model Overview of Loss Given Default LGD Models for Retail Credit LGD Models for Corporate Credit Economic Variables for LGD Estimation Estimating Downturn LGD Regression Techniques for LGD Ordinary Least Squares Linear Regression Ordinary Least Squares with Beta Transformation Beta Regression Ordinary Least Squares with Box-Cox Transformation Regression Trees Artificial Neural Networks Linear Regression and Non-linear Regression Logistic Regression and Non-linear Regression Performance Metrics for LGD Root Mean Squared Error Mean Absolute Error Area Under the Receiver Operating Curve Area Over the Regression Error Characteristic Curves R-square Pearson s Correlation Coefficient Spearman s Correlation Coefficient Kendall s Correlation Coefficient Model Development Motivation for LGD models Developing an LGD Model Case Study: Benchmarking Regression Algorithms for LGD Data Set Characteristics Experimental Set-Up Results and Discussion Chapter Summary References and Further Reading... 84

4 vi Chapter 5 Development of an Exposure at Default (EAD) Model Overview of Exposure at Default Time Horizons for CCF Data Preparation CCF Distribution Transformations Model Development Input Selection Model Methodology Performance Metrics Model Validation and Reporting Model Validation Reports Chapter Summary References and Further Reading Chapter 6 Stress Testing Overview of Stress Testing Purpose of Stress Testing Stress Testing Methods Sensitivity Testing Scenario Testing Regulatory Stress Testing Chapter Summary References and Further Reading Chapter 7 Producing Model Reports Surfacing Regulatory Reports Model Validation Model Performance Model Stability Model Calibration SAS Model Manager Examples Create a PD Report Create a LGD Report Chapter Summary

5 vii Tutorial A Getting Started with SAS Enterprise Miner A.1 Starting SAS Enterprise Miner A.2 Assigning a Library Location A.3 Defining a New Data Set Tutorial B Developing an Application Scorecard Model in SAS Enterprise Miner B.1 Overview B.1.1 Step 1 Import the XML Diagram B.1.2 Step 2 Define the Data Source B.1.3 Step 3 Visualize the Data B.1.4 Step 4 Partition the Data B.1.5 Step 5 Perform Screening and Grouping with Interactive Grouping B.1.6 Step 6 Create a Scorecard and Fit a Logistic Regression Model B.1.7 Step 7 Create a Rejected Data Source B.1.8 Step 8 Perform Reject Inference and Create an Augmented Data Set B.1.9 Step 9 Partition the Augmented Data Set into Training, Test and Validation Samples B.1.10 Step 10 Perform Univariate Characteristic Screening and Grouping on the Augmented Data Set B.1.11 Step 11 Fit a Logistic Regression Model and Score the Augmented Data Set B.2 Tutorial Summary Appendix A Data Used in This Book A.1 Data Used in This Book Chapter 3: Known Good Bad Data Chapter 3: Rejected Candidates Data Chapter 4: LGD Data Chapter 5: Exposure at Default Data Index From Developing Credit Risk Models Using SAS Enterprise Miner and SAS/STAT : Theory and Application, by Iain Brown. Copyright 2014, SAS Institute Inc., Cary, North Carolina, USA. ALL RIGHTS RESERVED.