Mine Cetinkaya-Rundel (Duke University, RStudio) & Debbie Yuster (Ramapo College)
Abstract
The tidymodels framework is a collection of R packages for modeling and machine learning using tidyverse principles. In this session, we'll introduce the tidymodels framework and discuss how it can be incorporated into an introductory data science curriculum. We will walk through case studies for building linear and logistic regression models with the goals of prediction and classification, performing cross-validation, and evaluating model performance. We will also compare the tidymodels approach to more traditional modeling frameworks in R as well as touch on how to do statistical inference within the tidymodels framework.
During the session, we will make use of polling questions and live coding. Participants will have the option to follow along with the exercises on RStudio Cloud.
The intended audience for the session is anyone who is interested in an introduction to tidymodels, either for their own use or in their teaching. The session will assume familiarity with R and basics of tidyverse (e.g. pipe operator, dplyr, ggplot2) as well as with modeling (linear and logistic regression).