siRSM (version 1.1)

siRSM: Single-index Response Surface Model

Description

This is the main function for users to call.

Usage

"siRSM"(y, U, V, trial, interaction.only=FALSE, use.parallel=TRUE)
"print"(x, ...)
"plot"(x, ...)

Arguments

y

response vector

U

matrix, whose columns are covariates for first factor

V

matrix, whose columns are covariates for second factor, must have same dimension as U

trial

number of different initial values to try --- if missing, defaults to 2*K, where K = ncol(U) = ncol(V)

interaction.only

fit an interaction-only (rather than a full-quadratic) model

use.parallel

if TRUE, tells underlying utility function multi.run to exploit multiple cores by using foreach, doSNOW, etc

x

an object of class siRSM, often result of siRSM

...

arguments for plot include: xname=character string, name for first (composite) factor; yname=character string, name for second (composite) factor; zname=character string, name for response; center=if 'zero', centers the surface plot at (0,0)

Value

siRSM. For most users, the most useful elements are:

w

the estimated single index vector

coef

the coefficients of the corresponding response surface