Variable Selection with Machine Learning Models: LASSO and Alternatives
presented by
Dr. Elizabeth Handorf
Associate Professor, Department of Biostatistics and Epidemiology and Resident Faculty, Rutgers Cancer Institute
4/24, 1 – 2:30 pm on Zoom
Target Audience: Students, statisticians, data scientists, and other researchers who are familiar with regression, and who are interested in learning about machine learning models.
Workshop Description: Large observational datasets, from US Census data, cancer registries, and electronic health records provide a rich source of information we can use to understand health outcomes. Analysis of these datasets can be challenging, as the large number of potential predictors makes traditional regression problematic. Machine learning models may present a better alternative. Many machine learning models support some type of variable selection, which helps us understand what variables are driving model predictions. In this workshop, we will discuss a few popular machine learning models which incorporate variable selection and see how well they perform. We will then discuss a clinical example, using these models to identify patients at risk of delays in starting cancer treatment. Finally, we will learn how to use one of these methods, the LASSO, with R software.
View Flyer and Register