CV_reg {OSCV}R Documentation

The cross-validation (CV) function in the regression context.

Description

Computing CV(h)CV(h), the value of the CV function in the regression context.

Usage

CV_reg(h, desx, y)

Arguments

h

numerical vector of bandwidth values,

desx

numerical vecror of design points,

y

numerical vecror of data values corresponding to the design points desxdesx.

Details

The CV function is a measure of fit of the regression estimate to the data. The local linear estimator based on the Gaussian kernel is used. The cross-validation bandwidth is the minimizer of the CV function.

Value

The vector of values of CV(h)CV(h) for the correponsing vector of hh values.

References

Stone, C.J. (1977) Consistent nonparametric regression. Annals of Statistics, 5(4), 595-645.

See Also

loclin, h_ASE_reg, ASE_reg, OSCV_reg.

Examples

## Not run: 
# Example (Old Faithful geyser). Take x=waiting time; y=eruption duration. The sample size n=272.
xdat=faithful[[2]]
ydat=faithful[[1]]
harray=seq(0.5,10,len=100)
cv=CV_reg(harray,xdat,ydat)
R=range(xdat)
h_cv=round(optimize(CV_reg,c(0.01,(R[2]-R[1]/4)),desx=xdat,y=ydat)$minimum,digits=4)
dev.new()
plot(harray,cv,'l',lwd=3,xlab="h",ylab="CV(h)",main="CV function for the Old Faithful 
geyser data", cex.lab=1.7,cex.axis=1.7,cex.main=1.5)
legend(6,0.155,legend="n=272",cex=1.8,bty="n")
legend(1,0.18,legend=paste("h_CV=",h_cv),cex=2,bty="n")

## End(Not run)

[Package OSCV version 1.0 Index]