model.extract
Extract Special Information from Model Frame
Description
Extracts Special Information from Model Frame
Usage
model.extract(frame, component)
model.response(data, type="any")
model.offset(x)
model.weights(x)
Arguments
| frame, data, x |
a model frame, the result of calling model.frame()
|
| component |
a character string that specifies the name of the information to
retrieve. Built-in requests include "response",
"weights", and three flavors of predictor matrix,
"tree.x", "loess.x", and "lm.x" (a synonym for
model.matrix()), for the use of the corresponding fitting
functions. Other names are wrapped in parentheses and selected as the
corresponding variable of the model frame, following the convention of
model.frame() for special= arguments.
|
| type |
a character string given the type of response returned.
Possible values are "any", "numeric", "double".
If "numeric" or "double", the results are
coerced with a call to as.numeric.
|
Value
returns the selected variable from the model frame. An important
special feature is that the names or row labels of the extracted data
are set to the row names of the model frame. This is the mechanism
that transmits row names to the fitted values and residuals in model-
fitting.
This function is normally a support routine for fitting functions,
such as lm().
Note
The model.response, model.weights, and
model.offset functions are wrapper functions that
model.extract uses to get the appropriate component.
The model.weights and model.offset
functions do not retain the row names of the model frame like model.extract.