fitMultiKernel: Multivariate kernel-machine regression

In turgeonmaxime/multiKernel: Multivariate prediction using kernel-machine regression

Description

This function performs multivariate kernel-machine regression by minimizing a specific loss function

Usage

fitMultiKernel(response, covariate, confounder = NULL, kernel = c("linear",
  "quadratic", "gaussian"), intercept = TRUE, tau, pure = FALSE, ...)

Arguments

response	matrix of response variables
covariate	matrix of covariate variables, which are included in the kernel.
confounder	matrix or data.frame of confounder variables, which are not included in the kernel.
kernel	Type of kernel to use.
intercept	Should we include an intercept?
tau	Tuning parameter.
pure	Logical. Use the pure R version?
...	Extra parameters to be passed to the kernel function.

Details

If confounder = NULL, intercept = FALSE, and response contains only one response variable, then this is equivalent to kernel ridge regression.

Value

Returns the kernel predictor.

turgeonmaxime/multiKernel documentation built on June 1, 2019, 2:56 a.m.