Variable Selection and Penalized Regression Using the REGSELECT Procedure

This workshop introduces the REGSELECT procedure in SAS® Visual Statistics software, showcasing its capabilities for variable selection and penalized regression. The procedure offers several penalized regression methods with both convex and nonconvex penalties, including LASSO, elastic net, SCAD, MCP, and so on. PROC REGSELECT provides extensive capabilities for customizing the model selection by using a wide variety of selection and stopping criteria, ranging from computationally efficient significance-level-based criteria to computationally intensive validation-based criteria. The workshop presents practical examples to illustrate software usage. It is intended for statisticians and data analysts who are interested in high-dimensional statistical modeling and modern statistical learning.