plus (version 1.0)

plot.plus: Plot method for plus objects

Description

Produce a plot from a plus solution path.

Usage

"plot"(x, xvar=c("lam","step"), yvar=c("coef","newy","lam","dim","R-sq"), 
	newx, step.interval, lam.interval, predictors, ...)

Arguments

x

a plus object

xvar

penalty level or plus step as the variable for the horizontal axis in the plot. Default is "lam".

yvar

paths of coefficients, predictions, penalty level, the number of nonzero coefficients or R-square as the variable for the vertical axis in the plot. Default is "coef"

newx

x values at which the fit is required. If newx is not set and yvar is "newy", no plot is produced.

step.interval

lower and upper bounds of the x-axis in the plot when xvar is "step". Defult covers all steps in the computed path.

lam.interval

lower and upper bounds of the x-axis in the plot when xvar is "lam". Default covers all penalty levels in the computed path.

predictors

a subset of predictors for which coefficients are plotted. Default is the entire set of predictors.

...

Additonal arguments for generic methods

Value

Details

The fitted coefficients and penalty levels are linear between two consecutive turning points in the plus path so that exact values of "coef", "newy" and "lam" are ploted when xvar is set as "step". For concave penalties, the solution path is not necessarily monotone in penalty level. Since the extracted coefficients for a particular given penalty level is defined as the first point at which the solution path hits the given penalty level, the "coef" and "newy" plotted as approximations as the linear interpolation of their actual values at specifiec lam when xvar is set as "lam".

References

Zhang, C.-H. (2010). Nearly unbiased variable selection under minimax concave penalty. Annals of Statistics 38, 894-942.

Examples

data(sp500)
attach(sp500)

x <- sp500.percent[,3: (dim(sp500.percent)[2])] 
y <- sp500.percent[,1]

object <- plus(x,y,method="mc+")
plot(object)
detach(sp500)