Statistical Issues in Predictive Modeling
Whereas Chapter 10 describes the basic process of predictive modeling, this chapter goes into depth on three key issues: selection of variables, treatment of missing data, and evaluation of models. Topics covered include stepwise selection and principal components methods of variable selection; imputation methods, missing variable dummies, and data fusion techniques for missing data; and validation techniques and metrics for evaluating predictive models.
KeywordsValidation Sample Lorenz Curve Principal Component Regression Default Probability Single Imputation
Unable to display preview. Download preview PDF.