Bspl.cv: Choose number of basis functions in least squares B-splines...

In gregorkb/spaddinf: Fit two-step estimator for inference in sparse high-dimensional additive models

Description

Choose number of basis functions in least squares B-splines via crossvalidation

Usage

Bspl.cv(Y, X, d.seq, n.folds, plot = FALSE)

Arguments

Y	a response vector (centered)
X	vector of covariate observations
d.seq	a sequence of candidate numbers of basis functions
n.folds	the number of crossvalidation folds
plot	a logical indicating whether to plot crossvalidation output

Value

the number of basis functions selected by crossvalidation

Examples

data <- data_gen(n = 200,
                 q = 50,
                 r = .9)

cv.d <- Bspl.cv(Y = data$Y.oracle[,1],
                X = data$X[,1],
                d.seq = 3:12,
                n.folds = 5,
                plot = TRUE)

oracle.Bspl.out <- oracle.Bspl(Y = data$Y.oracle[,1],
                               X = data$X[,1],
                               d = cv.d,
                               plot = TRUE)

gregorkb/spaddinf documentation built on July 23, 2021, 4:02 a.m.