| StatModel-class {modeltools} | R Documentation |
A class for unfitted statistical models.
Objects can be created by calls of the form new("StatModel", ...).
name:"character", the name of the
model.dpp:"function", a function for
data preprocessing (usually formula-based). fit:"function", a function for
fitting the model to data.predict:"function", a function for
computing predictions.capabilities:"StatModelCapabilities".signature(model = "StatModel", data = "ModelEnv"):
fit model to data.
This is an attempt to provide unified infra-structure for unfitted
statistical models. Basically, an unfitted model provides a function for
data pre-processing (dpp, think of generating design matrices),
a function for fitting the specified model to data (fit), and
a function for computing predictions (predict).
Examples for such unfitted models are provided by linearModel and
glinearModel which provide interfaces in the "StatModel" framework
to lm.fit and glm.fit, respectively. The functions
return objects of S3 class "linearModel" (inheriting from "lm") and
"glinearModel" (inheriting from "glm"), respectively. Some
methods for S3 generics such as predict, fitted, print
and model.matrix are provided to make use of the "StatModel"
structure. (Similarly, survReg provides an experimental interface to
survreg.)
### linear model example df <- data.frame(x = runif(10), y = rnorm(10)) mf <- dpp(linearModel, y ~ x, data = df) mylm <- fit(linearModel, mf) ### equivalent print(mylm) lm(y ~ x, data = df) ### predictions Predict(mylm, newdata = data.frame(x = runif(10)))