Machine Learning: In the Trenches - Brenda Gillespie and Chris Andrews, University of Michigan

This talk will cover several practical aspects of machine learning, including

• the feasibility of machine learning with correlated outcome data, such as paired eyes,
• assessing variable importance with LASSO or Elastic Net in the presence of multi-collinearity among the covariates
• the option of combining features from different machine learning methods to improve prediction.

About the Presenters

Brenda Gillespie is Associate Director of the University of Michigan unit, Consulting for Statistics, Computing and Analytics Research.  Over the years, Brenda has worked on many medical and environmental research studies, particularly related to kidney disease, eye diseases, and environmental toxins. SAS is her primary statistical software, and she is delighted that SAS has developed many new tools for machine learning and artificial intelligence.

Chris Andrews is a statistical consultant at the University of Michigan Center for Statistical Consultation and Research, focusing on SAS and R. He also serves as a statistician in the UM Ophthalmology and Visual Sciences department at the Medical School, and is an Intermittent Lecturer in Epidemiology.