Usage
cat_control(adapted.weights = TRUE, assured.intercept = TRUE,
tuning.criterion = "deviance", K = 5, lambda.upper = 10000,
lambda.accuracy = 0.01, standardize = FALSE, accuracy = 2,
digits = 4, c = 10^(-5), g = 0.5, epsilon = 10^(-5),
maxi = 250, steps = 25, p.ord.abs = TRUE, cv.refit = FALSE, ...)
Arguments
adapted.weights
logical; if TRUE, penalty terms are weighted adaptively, that is by inverse ML-estimates; set to FALSE, if ML-estimates do not exist/are to close to zero
assured.intercept
logical; shall a constant intercept remain in the model in any case?
tuning.criterion
loss criterion for cross-validation; one out of "deviance", "SSE" (for sum of squared errors)
K
integer; folds for cross-validation
lambda.upper
integer; upper bound for cross-validation of lambda
lambda.accuracy
numeric; how accurate shall lambda be cross-validated?; minimal absolute difference between two candidates for lambda
standardize
logical; if TRUE all metric covariates are standardized by their empirical variances
accuracy
integer; number of digits being compared when setting coefficents equal/to zero; must be 4 or less
digits
integer; number of digits for estimates
c
small, positive, real constant; needed for a approximation of the absolute value function in the PIRLS-algorithm, see Ulbricht (2010) g
step length parameter for the PIRLS-algorithm, see Ulbricht (2010); out of )0,1( epsilon
small, positive, real constant; termination criterion for the PIRLS-algorithm, see Ulbricht (2010) maxi
integer; maximal number of iterations in the fitting algorithm
steps
integer; tuning parameter for path-plotting; minimal number of estimates employed for path-plotting
p.ord.abs
logical; if TRUE coefficients and adjacent differences of the coefficients belonging to an ordinal covariate are penalized absolutely; if FALSE all differences and the first coefficient are penalized.
cv.refit
logical; if TRUE cross-validation is based on a refit of selected coefficients
...
further arguments passed to or from other methods