TidyTuesday: Modeling Interaction Effects using TidyModels

In this #TidyTuesday video, I go over interaction effects using linear models. I use Max Khun's book on Feature Engineering to explain the framework for approaching how to model interactions and when there are too many features where I briefly describe Hierarchy and Heredity principles. I then go on to use base R linear models to select relevant features and show tidy select tricks to select and rename columns quickly. I then use the #TidyModels package to create the pairwise interactions and tune the Main Effects (first stage) lasso regression model and an Interaction Effects (Second Stage) lasso regression model. Finally, I examine how workflows can help reduce clutter in the environment and how to extract coefficients from a trained model.

Connect with me on LinkedIn:
Code for this video:
Max Khun's Book:

Rstudio,Tidyverse,R Programming,Data Science,Analytics,University of Iowa,Statistics,ggplot2,ggplot,data wrangling,Dplyr,Data Visualizaiton,Data Viz,EDA,TidyTuesday,RStats,TidyModels,Caret,ML,Machine Learning,Data,Data Modeling,Black-Box,Recipes,TidyModels Package,CRISP-DM,Lasso Regression,Interaction Effects,Linear Modeling,Linear Models,Recipe Package,Two-Stage Modeling,Penalized Regression,Partial Lease Squares,Parsnip Package,Grid Search,Regression,