2015 QPRC

Variable Selection via Penalized Regression: An Overview
Howard Bondell, North Carolina State University

Over the past two decades, penalized regressions have become popular methods to simultaneously estimate parameters and perform variable selection. These approaches have become particularly useful in the age of big data. In this talk, we will discuss some of the main ideas behind these shrinkage methods. Both the motivation and implementation of penalized regressions will be discussed for linear regression. The main focus will be on the Least Absolute Shrinkage and Selection Operator (LASSO), but extensions to various other penalties will also be discussed.