This documents the methods that ref_grid calls. A user
or package developer may add emmeans support for a model
class by writing recover_data and emm_basis methods
for that class. (Users in need for a quick way to obtain results for a model
that is not supported may be better served by the qdrg function.)
recover_data(object, ...)# S3 method for call
recover_data(object, trms, na.action, data = NULL, params = "pi", frame, ...)
emm_basis(object, trms, xlev, grid, ...)
.recover_data(object, ...)
.emm_basis(object, trms, xlev, grid, ...)
.emm_register(classes, pkgname)
An object of the same class as is supported by a new method.
Additional parameters that may be supported by the method.
The terms component of object (typically with
the response deleted, e.g. via delete.response)
Integer vector of indices of observations to ignore; or
NULL if none
Data frame. Usually, this is NULL. However, if non-null,
this is used in place of the reconstructed dataset. It must have all of the
predictors used in the model, and any factor levels must match those used
in fitting the model.
Character vector giving the names of any variables in the model
formula that are not predictors. For example, a spline model may involve
a local variable knots that is not a predictor, but its value is
needed to fit the model. Names of parameters not actually used are harmless,
and the default value "pi" (the only numeric constant in base R)
is provided in case the model involves it. An example involving splines
may be found at https://github.com/rvlenth/emmeans/issues/180.
Optional data.frame. Many model objects contain the
model frame used when fitting the model. In cases where there are no
predictor transformations, this model frame has all the original predictor
values and so is usable for recovering the data. Thus, if frame is
non-missing and data is NULL, a check is made on trms
and if there are no function calls, we use data = frame. This
can be helpful because it provides a modicum of security against the
possibility that the original data used when fitting the model has been
altered or removed.
Named list of factor levels (excluding ones coerced to factors in the model formula)
A data.frame (provided by ref_grid) containing
the predictor settings needed in the reference grid
Character names of one or more classes to be registered.
The package must contain the functions recover_data.foo and
emm_basis.foo for each class foo listed in classes.
Character name of package providing the methods (usually
should be the second argument of .onLoad)
The recover_data method must return a data.frame
containing all the variables that appear as predictors in the model,
and attributes "call", "terms", "predictors",
and "responses". (recover_data.call will
provide these attributes.)
The emm_basis method should return a list with the
following elements:
The matrix of linear functions over grid, having the same
number of rows as grid and the number of columns equal to the length
of bhat.
The vector of regression coefficients for fixed effects. This
should include any NAs that result from rank deficiencies.
A matrix whose columns form a basis for non-estimable functions
of beta, or a 1x1 matrix of NA if there is no rank deficiency.
The estimated covariance matrix of bhat.
A function of (k, dfargs) that returns the degrees of
freedom associated with sum(k * bhat).
A list containing additional arguments needed for
dffun
.recover_data and .emm_basis are hidden exported versions of
recover_data and emm_basis, respectively. They run in emmeans's
namespace, thus providing access to all existing methods.
If the recover_data method generates information needed by emm_basis,
that information may be incorporated by creating a "misc" attribute in the
returned recovered data. That information is then passed as the misc
argument when ref_grid calls emm_basis.
Some models may need something other than standard linear estimates and
standard errors. If so, custom functions may be pointed to via the items
misc$estHook, misc$vcovHook and misc$postGridHook. If
just the name of the hook function is provided as a character string, then it
is retrieved using get.
The estHook function should have arguments (object, do.se, tol,
...) where object is the emmGrid object,
do.se is a logical flag for whether to return the standard error, and
tol is the tolerance for assessing estimability. It should return a
matrix with 3 columns: the estimates, standard errors (NA when
do.se==FALSE), and degrees of freedom (NA for asymptotic). The
number of rows should be the same as object@linfct. The
vcovHook function should have arguments (object, tol, ...) as
described. It should return the covariance matrix for the estimates. Finally,
postGridHook, if present, is called at the very end of
ref_grid; it takes one argument, the constructed object, and
should return a suitably modified emmGrid object.
The .emm_register function is provided as a convenience to conditionally
register your
S3 methods for a model class, recover_data.foo and emm_basis.foo,
where foo is the class name. Your package should implement an
.onLoad function and call .emm_register if emmeans is
installed. See the example.
To create a reference grid, the ref_grid function needs to reconstruct
the data used in fitting the model, and then obtain a matrix of linear
functions of the regression coefficients for a given grid of predictor
values. These tasks are performed by calls to recover_data and
emm_basis respectively. A vignette giving details and examples
is available via vignette("xtending", "emmeans")
To extend emmeans's support to additional model types, one need only
write S3 methods for these two functions. The existing methods serve as
helpful guidance for writing new ones. Most of the work for
recover_data can be done by its method for class "call",
providing the terms component and na.action data as additional
arguments. Writing an emm_basis method is more involved, but the
existing methods (e.g., emmeans:::emm_basis.lm) can serve as models.
Certain recover_data and emm_basis methods are exported from
emmeans. (To find out, do methods("recover_data").) If your
object is based on another model-fitting object, it
may be that all that is needed is to call one of these exported methods and
perhaps make modifications to the results. Contact the developer if you need
others of these exported.
If the model has a multivariate response, bhat needs to be
“flattened” into a single vector, and X and V must be
constructed consistently.
In models where a non-full-rank result is possible (often, you can tell by
seeing if there is a singular.ok argument in the model-fitting
function), summary.emmGrid and its relatives check the
estimability of each
prediction, using the nonest.basis function in
the estimability package.
The models already supported are detailed in the "models" vignette. Some packages may provide additional emmeans support for its object classes.
# NOT RUN {
#--- If your package provides recover_data and emm_grid methods for class 'mymod',
#--- put something like this in your package code -- say in zzz.R:
.onLoad = function(libname, pkgname) {
if (requireNamespace("emmeans", quietly = TRUE))
emmeans::.emm_register("mymod", pkgname)
}
# }
Run the code above in your browser using DataLab