| varNPreg {lokern} | R Documentation |
Estimates the error variance σ^2 nonparametrically in the model
Y_i = m(x_i) + E_i,
where E_i ~ (0,sigma^2), i.i.d.
Computes leave-one-out residuals (local linear approximation followed by reweighting) and their variance.
varNPreg(x, y)
x |
abscissae values, ordered increasingly. |
y |
observations at y[i] at x[i]. |
A list with components
res |
numeric; residuals at x[] of length n. |
snr |
explained variance of the true curve |
sigma2 |
estimation of residual variance, sigma^2. |
This is an R interface to the resest Fortran subroutine, used
in lokerns and glkerns, see their help
pages for references and context.
Earlier version of the lokern package accidentally, contained
varest() which has been an identical copy of varNPreg().
Martin Maechler
x <- sort(runif(100)) y <- sin(pi*x) + rnorm(100)/10 str(ve <- varNPreg(x,y))