Week 5: Linear Models Part 2 (James Scott)

James Scott will be speaking to us about logit and Poisson models on Friday, September 27th. James is an Assistant Professor of Statistics affiliated with both the Department of information, Risk, and Operations Management and the Division of Statistics and Scientific Computation. You can read more about his work on his website, http://jgscott.github.io/

You can download R scripts and datasets from GitHub. You'll want the files contained in folders '03logit' and '04counts'.  https://github.com/jgscott/mdar

The R scripts require the use of the following packages, which you can download from CRAN: 'mosaic', 'multcomp', 'nnet', 'lmtest', 'MASS', 'contrast', and 'lattice'. 

Summary notes on multivariate analysis: The first two sections of this document recap the concepts we discussed last week in Nate Pope's intro to linear models. James will discuss content from sections 3 and 4 of this document, which deal with logistic regression (section 3) and models for count outcomes (that is, Poisson models) (section 4).  SummaryNotes 

Further notes on logit models: http://jgscott.github.io/SSC325/files/05-DiscreteOutcomes.pdf