Credit scoring, response modeling, and insurance rating : a practical guide to forecasting consumer behavior

Online-Ressource (xvii, 297 p)

Includes bibliographical references and index

Cover; Contents; 1 Introduction; 1.1 Scope and content; 1.2 Model applications; 1.3 The nature and form of consumer behavior models; 1.3.1 Linear models; 1.3.2 Classification and regression trees (CART); 1.3.3 Artificial neural networks; 1.4 Model construction; 1.5 Measures of performance; 1.6 The stages of a model development project; 1.7 Chapter summary; 2 Project Planning; 2.1 Roles and responsibilities; 2.2 Business objectives and project scope; 2.2.1 Project scope; 2.2.2 Cheap, quick or optimal?; 2.3 Modeling objectives; 2.3.1 Modeling objectives for classification models

2.3.2 Roll rate analysis2.3.3 Profit based good/bad definitions; 2.3.4 Continuous modeling objectives; 2.3.5 Product level or customer level forecasting?; 2.4 Forecast horizon (outcome period); 2.4.1 Bad rate (emergence) curves; 2.4.2 Revenue/loss/value curves; 2.5 Legal and ethical issues; 2.6 Data sources and predictor variables; 2.7 Resource planning; 2.7.1 Costs; 2.7.2 Project plan; 2.8 Risks and issues; 2.9 Documentation and reporting; 2.9.1 Project requirements document; 2.9.2 Interim documentation; 2.9.3 Final project documentation (documentation manual); 2.10 Chapter summary

3 Sample Selection3.1 Sample window (sample period); 3.2 Sample size; 3.2.1 Stratified random sampling; 3.2.2 Adaptive sampling; 3.3 Development and holdout samples; 3.4 Out-of-time and recent samples; 3.5 Multi-segment (sub-population) sampling; 3.6 Balancing; 3.7 Non-performance; 3.8 Exclusions; 3.9 Population flow (waterfall) diagram; 3.10 Chapter summary; 4 Gathering and Preparing Data; 4.1 Gathering data; 4.1.1 Mismatches; 4.1.2 Sample first or gather first?; 4.1.3 Basic data checks; 4.2 Cleaning and preparing data; 4.2.1 Dealing with missing, corrupt and invalid data

4.2.2 Creating derived variables4.2.3 Outliers; 4.2.4 Inconsistent coding schema; 4.2.5 Coding of the dependent variable (modeling objective); 4.2.6 The final data set; 4.3 Familiarization with the data; 4.4 Chapter summary; 5 Understanding Relationships in Data; 5.1 Fine classed univariate (characteristic) analysis; 5.2 Measures of association; 5.2.1 Information value; 5.2.2 Chi-squared statistic; 5.2.3 Efficiency (GINI coefficient); 5.2.4 Correlation; 5.3 Alternative methods for classing interval variables; 5.3.1 Automated segmentation procedures

5.3.2 The application of expert opinion to interval definitions5.4 Correlation between predictor variables; 5.5 Interaction variables; 5.6 Preliminary variable selection; 5.7 Chapter summary; 6 Data Transformation (Pre-processing); 6.1 Dummy variable transformed variables; 6.2 Weights of evidence transformed variables; 6.3 Coarse classing; 6.3.1 Coarse classing categorical variables; 6.3.2 Coarse classing ordinal and interval variables; 6.3.3 How many coarse classed intervals should there be?; 6.3.4 Balancing issues; 6.3.5 Applying transformations to holdout, out-of-time and recent samples

6.4 Which is best - weight of evidence or dummy variables?

Electronic reproduction; Available via World Wide Web

10.1057/9781137031693 [DOI]

ECONIS - Online Catalogue of the ZBW