EDR (version 0.6-7)

predict.edr: Predict values of the link function for a fitted edr-object

Description

The function allows to predict values of the link function in a multi-index model with estimated effective dimension reduction space.

Usage

# S3 method for edr
predict(object, xest, m = 1, h=NULL, method = "sm",...)

Arguments

object

object of class edr

xest

matrix of design points where values of the link function are to be predicted

m

specified dimension of the dimension reduction space.

h

bandwidth, if h=NULL a bandwidth is guessed from the estimation

method

method="sm" requires and uses package sm while method="direct" specifies a local linear smoother.

...

Additional parameters that are currently not evaluated.

Value

List with components.

x

The content of argument xest

fhat

Predicted values of the link function

Details

This function provides predictions of response values based on a multiindex model analyzed using function edr. It requires specification of the dimension of the dimension reduction space in argument m. Argument method allows to select between two different local smoothers for estimation of values of the link function. The use of method = "sm" requires package sm and is restricted to m<=2, see documentation of package sm. method = "direct" uses an implementation of a local linear smoother. If parameters of the local linear smoother are not identified for a given bandwidth h a kernel smoother or if this is not identified a 1-nearest neighbor estimate is used. The bandwidth is, in case of h=NULL, is determined by generalized cross-validation for sm and as x$h*x$rho in case of the local linear smoother.

References

M. Hristache, A. Juditsky, J. Polzehl and V. Spokoiny (2001). Structure adaptive approach for dimension reduction, The Annals of Statistics. Vol.29, pp. 1537-1566.

J. Polzehl, S. Sperlich (2008). A Note on Stuctural Adaptive Dimension Reduction, Journal of Statistical Computation and Simulation, DOI: 10.1080/00949650801959699

See Also

edr,plot.edr, summary.edr, print.edr, edr.R, edrcv

Examples

# NOT RUN {
require(EDR)
# }
# NOT RUN {
demo(edr_ex4)
# }