lqa.control {lqa} | R Documentation |
Auxiliary function as user interface for lqa
fitting. Typically only used when calling lqa
or lqa.update2
.
lqa.control (x = NULL, var.eps = .Machine$double.eps, max.steps = 5000, conv.eps = 0.001, conv.stop = TRUE, c1 = 1e-08, digits = 5, ...)
x |
object of class 'lqa'. This optional argument is just included to be in line with the S3 class concept. |
var.eps |
tolerance in checking for zero variance of some regressors. |
max.steps |
maximum number of steps in the lqa algorithm. |
conv.eps |
tolerance for convergence break in parameter updating. |
conv.stop |
whether or not to stop the iterations when estimated coefficients are converged. |
c1 |
controls the amount of approximation of linear combinations in the penalty term. |
digits |
number of digits of tuning parameter candidates to take into consideration when returning the loss array and mean
array in |
... |
further arguments. |
A list with the arguments as components.
Jan Ulbricht
set.seed (1111) n <- 200 p <- 5 X <- matrix (rnorm (n * p), ncol = p) X[,2] <- X[,1] + rnorm (n, sd = 0.1) X[,3] <- X[,1] + rnorm (n, sd = 0.1) true.beta <- c (1, 2, 0, 0, -1) y <- drop (X %*% true.beta) + rnorm (n) control.obj <- lqa.control (max.steps = 200, conv.eps = 1e-3, conv.stop = FALSE) obj <- lqa (y ~ X, family = gaussian (), penalty = lasso (1.5), control = control.obj) obj$coef