R/plot.cv.creNet.R

In kouroshz/creNet: Fit a GLM with a combination of lasso and overlapping group lasso regularization

plot.cv.creNet = function(x,...){
  	cvobj <- x
	cvobj$lldiff <- cvobj$lldiff[[cvobj$best.alpha[1]]]
	cvobj$llSD <- cvobj$llSD[[cvobj$best.alpha[1]]]
	cvobj$lambdas <- cvobj$lambdas[[cvobj$best.alpha[1]]]
  	xlab = "log(Lambda)"
  	ylab = "CV Negative Log Likelihood"
  	cvUp <- cvobj$lldiff + cvobj$llSD
  	cvDn <- cvobj$lldiff - cvobj$llSD
  	plot.args = list(x=log(cvobj$lambdas), y = cvobj$lldiff, xlab = xlab, ylab = ylab, ylim = range(cvUp, cvDn))
  	do.call("plot", plot.args)
  	error.bars(log(cvobj$lambdas), cvUp, cvDn, width = 0.01, col ="darkgrey")
  	points(log(cvobj$lambdas),cvobj$lldiff,pch=20,col="red")
}