Regression Modeling Strategies

3 credits · Graduate School

This course presents strategies for, and a survey of current thinking on, building multivariable regression models primarily for the purpose of prediction, but also for estimation and inference. Topics include using regression splines to relax linearity assumptions, the perils of variable selection and over-fitting, where to spend degrees of freedom, shrinkage, imputation of missing data, data reduction, and interaction surfaces. There is also an emphasis placed on describing approaches for graphically understanding models and using resampling to estimate a model’s likely performance on new data. Statistical methods related to binary logistic models, ordinal logistic models, and survival models are covered. Students will develop, validate, and graphically describe multivariable regression models. Prerequisites: BIOS 6312 or equivalent. [3]