vcrpart (version 1.0-6)

tvcm-methods: Methods for tvcm objects

Description

Standard methods for computing on tvcm objects.

Usage

# S3 method for tvcm
coef(object, ...)
# S3 method for tvcm
depth(x, root = FALSE, ...)
# S3 method for tvcm
extract(object, what = c(
              "control", "model", 
              "nodes", "sctest", "p.value",
              "devgrid", "cv", "selected",
              "coef", "sd", "var"),
        steps = NULL, ...)
# S3 method for tvcm
neglogLik2(object, ...)
# S3 method for tvcm
predict(object, newdata = NULL,
        type = c("link", "response", "prob", "class",
          "node", "coef", "ranef"),
        ranef = FALSE, na.action = na.pass, ...)
# S3 method for tvcm
splitpath(tree, steps = 1L,
         details = FALSE, ...)
# S3 method for tvcm
summary(object, ...)
# S3 method for tvcm
width(x, ...)

Value

The coef.tvcm and coefficients.tvcm

methods return a list with model coefficients. Slot vc stores varying coefficients, fe

fixed coefficients and re coefficients on random effects.

The depth.tvcm method returns a integer vector with the depth of the trees of every varying coefficient. width.tvcm returns a integer vector with the width of the trees.

The extract and extract.tvcm methods allow to extract further information of tvcm objects, such as the underlying regression model. The type of the return value depends on the input for argument what.

The formula.tvcm method extracts the model formula, which is an object of class formula. See also

formula.

The methods fitted.tvcm and

predict.fvcm return an object of class numeric

or matrix, depending on the used model or the specification of the argument type.

The getCall.tvcm method extracts the call for fitting the model, which is an object of class call. See also

call.

The logLik.tvcm method returns an object of class

logLik, see also logLik.

The model.frame.tvcm method returns a

data.frame. See also model.frame.

The neglogLik2.tvcm method returns a single numeric, see also neglogLik2.

The nobs.tvcm method extracts the number of observations used to fit the model. See also

nobs.tvcm.

The print.tvcm and summary.tvcm

methods return NULL.

The ranef.tvcm method returns an object of class

matrix with values for the random effects. See also

ranef.olmm and ranef.

The resid.tvcm and residuals.tvcm

methods return a numeric or a matrix, depending on the used model or the type of residuals. See the help of the

resid method of the used model.

The methods splitpath and

splitpath.tvcm return an object of class

splitpath.tvcm that contains information on splitting when building the tree.

The weights.tvcm method extracts a numeric

vector with the model weights. See also weights.

Arguments

object, tree, x

an object of class tvcm.

root

logical scalar. Should the root count be counted in depth?

steps

integer vector. The iteration steps from which information should be extracted.

newdata

an optional data frame in which to look for variables with which to predict, if omitted, the fitted values are used.

type

character string. Denotes for predict the type of predicted value. See predict.glm or predict.olmm. "response" and "prob" are identical.

na.action

function determining what should be done with missing values for fixed effects in newdata. The default is to predict NA: see na.pass.

ranef

logical scalar or matrix indicating whether prediction should be based on random effects. See predict.olmm.

what

a character specifying the quantities to extract.

details

logical scalar. Whether detail results like coefficient constancy tests or loss minimizing grid search should be shown.

...

Additional arguments passed to the calls.

Author

Reto Burgin

Details

The predict function has two additional options for the type argument. The option "node" calls the node id and "coef" predicts the coefficients corresponding to an observation. In cases of multiple vc terms for the same predictor, the coefficients are summed up.

The splitpath function allows to backtrack the partitioning procedure. By default, it shows which split was chosen in the first iteration. The interesting iteration(s) can be selected by the steps argument. With details = TRUE it is also possible to backtrack the coefficient constancy tests and/or the loss reduction statistics.

summary computes summary statistics of the fitted model, including the estimated coefficients. The varying coefficient are printed by means of a printed decision tree. Notice that in cases there is no split for the varying coefficient, the average coefficient will be among the fixed effects.

Further undocumented, available methods are: fitted, formula, getCall, logLik, model.frame, nobs, print, ranef, resid, and weights. All these methods have the same arguments as the corresponding default methods.

See Also

tvcm, tvcm-assessment, tvcm-plot

Examples

## ------------------------------------------------------------------- #
## Dummy example:
##
## Apply various methods on a 'tvcm' object fitted on the 'vcrpart_2'
## data. Cross-validation is omitted to accelerate the computations.
## ------------------------------------------------------------------- #

data(vcrpart_2)

model <- tvcm(y ~ -1 + vc(z1, z2) + vc(z1, z2, by = x1) + x2,
              data = vcrpart_2, family = gaussian(), subset = 1:90,
              control = tvcm_control(cv = FALSE))

coef(model)
extract(model, "selected")
extract(model, "model")
predict(model, newdata = vcrpart_2[91:100,], type = "node")
predict(model, newdata = vcrpart_2[91:100,], type = "response")
splitpath(model, steps = 1)
summary(model, digits = 2)