pblm.control: Auxiliary for controlling the algorithm in a 'pblm' model

In pblm: Bivariate Additive Marginal Regression for Categorical Responses

Auxiliary for controlling the algorithm in a pblm model

Description

This is an auxiliary function for controlling the algorithm in a pblm model.

Usage

pblm.control(maxit = 30, maxit2 = 200, acc = 1e-07, acc2 = 1e-06, 
            zero.adj = 1e-06, l = NULL, restore.l = FALSE,    
            min.step.l = 1e-04, auto.select = FALSE, gaic.m = 2, 
            rss.tol = 1e-06, max.backfitting = 10, pgtol.df = 0.01, 
            factr.df = 1e+07, lmm.df = 5, parscale.df = 1, 
            max.gaic.iter = 500, pgtol.gaic = 1e-05, grad.tol = 1e-07, 
            factr.gaic = 1e+07, lmm.gaic = 5, parscale = 1, 
            conv.crit = c("dev", "pdev"))

Arguments

maxit	maximum number of Fisher-scoring iterations.
maxit2	maximum number of Newton-Raphson iterations for the inversion \eta->\pi.
acc	tolerance to be used for the estimation.
acc2	tolerance to be used for the inversion \eta->\pi.
zero.adj	adjustment factor for zeros in the probability vector \pi.
l	numerical, ranged in (0,1], representing the initial value of step lenght. By default l=1.
restore.l	logical, should the step length be restored to its initial value after each iteration? This is an experimental option and may be changed in the future.
min.step.l	numerical, minimum value fixed for the step length.
auto.select	logical, should the smoothing parameters be estimated by GAIC minimization? If TRUE The optimization will be performed numerically by using optim.
gaic.m	the "penalty" per parameter of the generalized AIC. By default it is 2, corresponding to the classical AIC.
rss.tol	tolerance for the residual sum of squares used in the backfitting algorithm.
max.backfitting	maximum number of backfitting iterations.
pgtol.df	tolerance to be used in order to get an amount of smoothing corresponding to the fixed degrees of freedom for the additive part. See argument pgtol from optim.
factr.df	numerical. For degrees-of-freedom optimization in the additive part. See argument factr from optim.
lmm.df	integer. For degrees-of-freedom optimization in the additive part. See argument lmm from optim.
parscale.df	A vector of scaling parameters for vector lambda when optimizing lambda for fixed degrees of freedom. See argument parscale from optim.
max.gaic.iter	integer. Maximum number of iterations for automatic model optimization. See argument maxit from optim.
pgtol.gaic	numerical. Tolerance to be used for automatic selection of smoothing parameters. See argument pgtol from optim.
grad.tol	numerical. Tolerance to be used when inverting the gradient matrix.
factr.gaic	numerical. For automatic selection of smoothing parameters. See argument factr from optim.
lmm.gaic	integer. For automatic selection of smoothing parameters. See argument lmm from optim.
parscale	A vector of scaling parameters for vector lambda for automatic model optimization. See argument parscale from optim.
conv.crit	Convergence criterion for model estimation. The default is "dev", corresponding to log-likelihood maximization. Alternatively, "pdev" is concerned with maximum penalized log-likelihood.

Value

A list with the same arguments of the function, unless unlikely specified by the user.

Author(s)

Marco Enea

See Also

pblm

pblm documentation built on June 19, 2025, 5:08 p.m.