Logistic Regression: Basics and Beyond

This class presents light theory, supported by simulations, as well as practical suggestions for understanding and developing binary logistic regression models. Topics include: Binary response model overview, the logistic regression model as implemented by SAS® PROC LOGISTIC, the likelihood function, properties of predictor and model fit statistics, Firth method versus maximum likelihood method, screening, binning, transforming of nominal, ordinal, discrete, and continuous predictors, identification of multicollinearity, oversampling for rare events, predictor variable selection methods using PROC LOGISTIC, HPLOGISTIC, HPGENSELECT including Best Subsets, SBC, Lasso, and Validation methods, and measures of fit and predictive accuracy including c statistic, KS statistic, classification error, and lift charts in the context of training, cross-validation and validation samples. Base SAS and SAS/Stat were used in the development of this class. There is no requirement for experience with Enterprise Guide, Enterprise Miner or Visual Analytics.

I will provide:
Printed copy of PowerPoint slides, printed supplementary course material, flash drive loaded with SAS programs

Prerequisites
Intermediate programming skill in base SAS. Some practical knowledge in using logistic regression or one college course in statistical modeling.

About the Presenter

Bruce Lund is an independent consultant who specializes in predictive modeling. For the prior 15 years he was a consultant for OneMagnify of Detroit. Before OneMagnify, he was the customer database manager at Ford Motor Company and a mathematics professor at University of New Brunswick, Canada. He has a mathematics PhD from Stanford University. Bruce has presented at SAS Global Forum, SAS AnalyticsX, ASA CSP, and frequently at MWSUG and MSUG. He has used SAS for over 30 years.