CV: Cross Validation for finding RKHS smoothing mean parameters

In sxz155/PFDA:

Description

This function apply two "Regular" and "Irregular" methods to find the penalty (φ) and kernel range parameter (ρ) in an RKHS smoothing mean

Usage

CV(grid, Data, alpha = 1, beta = 0.1, kernel = "Exp", phi = c(0.01),
  ro = c(0.2), fold = 10, cv.penalty = "Regular", Rep = 1000,
  print.cv = TRUE, print.estimated.time = TRUE, col.drop = TRUE)

Arguments

grid	grid (x-axis) for each curve, default is equally espaced between 0 and 1.
Data	a matrix which the of interest curves are located in columns
alpha, beta	Privacy parameters, real numbers
kernel	kernel function, can be "Exp" (Exponential kernel), "M3/2" (Matern precess with ν=3/2) "M5/2" (Matern precess with ν=5/2) "Gau" (Gaussian kernel) and "Sob" (Sobolev kernel) else define it as a bivariate kernel function with parameters "t" and "s" and a range parameter "ro".
phi	a real vector of penalty parameters, It will be done a grid search on them to find the minimum Cross Validation
ro	a real vector of kernel range parameters, It will be done a grid search on them to find the minimum Cross Validation
fold	number of fold using in Cross Validation
cv.penalty	"Regular" or "Irregular"
Rep	number of replications, (just) for "Irregular" method

Value

par: the optimum penalty (φ) and kernel range parameter (ρ) in an RKHS smoothing mean which gives the minimum Cross Validation

time: estimatation of remaining time

sxz155/PFDA documentation built on May 30, 2019, 10:40 p.m.