mlt-methods: Methods for mlt Objects

In mlt: Most Likely Transformations

Methods for mlt Objects

Description

Methods for objects of class mlt

Usage

## S3 method for class 'mlt'
coef(object, fixed = TRUE, ...)
coef(object) <- value
## S3 method for class 'mlt'
weights(object, ...)
## S3 method for class 'mlt'
logLik(object, parm = coef(object, fixed = FALSE), w = NULL, newdata, ...)
## S3 method for class 'mlt'
vcov(object, parm = coef(object, fixed = FALSE), complete = FALSE, ...)
Hessian(object, ...)
## S3 method for class 'mlt'
Hessian(object, parm = coef(object, fixed = FALSE), ...)
Gradient(object, ...)
## S3 method for class 'mlt'
Gradient(object, parm = coef(object, fixed = FALSE), ...)
## S3 method for class 'mlt'
estfun(x, parm = coef(x, fixed = FALSE),
       w = NULL, newdata, ...)
## S3 method for class 'mlt'
residuals(object, parm = coef(object, fixed = FALSE), 
       w = NULL, newdata, what = c("shifting", "scaling"), ...)
## S3 method for class 'mlt'
mkgrid(object, n, ...)
## S3 method for class 'mlt'
bounds(object)
## S3 method for class 'mlt'
variable.names(object, ...)
## S3 method for class 'mlt_fit'
update(object, weights = stats::weights(object), 
       subset = NULL, offset = object$offset, theta = coef(object, fixed = FALSE), 
       fixed = NULL, ...)
## S3 method for class 'mlt'
as.mlt(object)

Arguments

object, x	a fitted conditional transformation model as returned by mlt
fixed	a logical indicating if only estimated coefficients (fixed = FALSE) should be returned OR (for update) a named vector of fixed regression coefficients; the names need to correspond to column names of the design matrix
value	coefficients to be assigned to the model
parm	model parameters
w	model weights
what	type of residual: shifting means score with respect to a constant intercept for the shift term and scaling means score with respect to a constant intercept in the scaling term. This works whether or not such terms are actually present in the model
weights	model weights
newdata	an optional data frame of new observations. Allows evaluation of the log-likelihood for a given model object on these new observations. The parameters parm and w are ignored in this situation.
n	number of grid points
subset	an optional integer vector indicating the subset of observations to be used for fitting.
offset	an optional vector of offset values
theta	optional starting values for the model parameters
complete	currently ignored
...	additional arguments

Details

coef can be used to get and set model parameters, weights and logLik extract weights and evaluate the log-likelihood (also for parameters other than the maximum likelihood estimate). Hessian returns the Hessian (of the negative log-likelihood) and vcov the inverse thereof. Gradient gives the negative gradient (minus sum of the score contributions) and estfun the negative score contribution by each observation. mkgrid generates a grid of all variables (as returned by variable.names) in the model. update allows refitting the model with alternative weights and potentially different starting values. bounds gets bounds for bounded variables in the model.

mlt documentation built on April 11, 2025, 3:09 p.m.