MCVIS: A New Framework for Collinearity Discovery, Diagnostic, and Visualization
Chen Lin, Kevin Wang, Samuel Mueller
Abstract
Collinearity discovery through diagnostic tools is an important analysis step when performing linear regression. Despite their wide-spread use, collinearity indices such as the variance inflation factor and the condition number have limitations and may not be effective in some applications. In this article, we will contribute to the study of conventional collinearity indices through theoretical and empirical work. We will present mcvis, a new framework that uses resampling techniques to repeatedly learn from these conventional collinearity indices to better understand the causes of collinearity. Our framework is made available in R through the mcvis package which includes new collinearity measures and visualizations, in particular a bipartite plot that informs on the degree and structure of collinearity. Supplementary materials for this article are available online.