Extract 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
the name of the information desired. Built-in requests include "response", "weights", and three flavors of predictor matrix, "tree.x", "loess.x", and "lm.x", for the use of the corresponding fitting functions. The last is just a synonym for model.matrix(). 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:

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 is normally a support routine for fitting functions, such as lm().

The model.response, model.weights and model.offset functions are wrapper functions to model.extract to get the appropriate component. The model.weights and model.offset functions do not retain the row names of the model frame like the model.extract call does.