IRCcheck: Irrepresentable Condition Check
Check the irrepresentable condition (IRC) in both L1-regularized regression <doi:10.1109/TIT.2006.883611>
and Gaussian graphical models. The IRC requires that the important and unimportant variables
are not correlated, at least not all that much, and it is necessary for consistent model
selection. Exploring the IRC as a function of the number of variables, assumed sparsity,
and effect size can provide valuable insights into the model selection
properties of L1-regularization.
Please use the canonical form
to link to this page.