predict
Make Predictions from a Fitted Model Object
Description
Returns a vector or array of predictions.
This function is an S Version 3 generic (see Methods);
method functions can be written to handle specific
S Version 3 classes of data. Classes that already
have methods for this function include:
predict.bs,
predict.censorReg,
predict.coxph.penal,
predict.coxph,
predict.discrim,
predict.factanal,
predict.gam,
predict.glm,
predict.gls,
predict.gnls,
predict.lm,
predict.lmList,
predict.lmRobMM,
predict.lme,
predict.loess,
predict.mlm,
predict.nlme,
predict.nls,
predict.ns,
predict.princomp,
predict.princomp,
predict.smooth.spline.fit,
predict.smooth.spline,
predict.survReg.penal,
predict.survReg,
predict.survreg,
predict.tree.
Usage
predict(object, ...)
predict(object, newdata, ...) # most methods have second argument newdata
predict(object, newx, ...) # some methods have second argument newx
Arguments
| object |
a fitted model object.
|
| newdata, newx |
Most methods have an optional second argument named
newdata or
newx
indicating new data (explanatory variables) for which predictions
are desired. If omitted then predictions (fitted values) for
the original data are returned.
|
Value
depends on the method; typically
a vector or array of predictions, a list consisting of the
predictions and their standard errors,
or a list containing x and
y components of points on a prediction curve.
A standard use of predict is to simply extract the fitted values
from a fit object, or in the case of generalized models,
to extract the linear or additive predictor.
Warning
Calling predict on objects of class "lm" or "glm"
can produce incorrect predictions
when the newdata argument is used
if the formula in object involves data-dependent transformations,
such as poly(Age, 3) or sqrt(Age - min(Age)).
The predict.gam method overcomes this for the gam, glm,
and lm classes.
In other cases, this can be overcome by explicitly supplying the
derived matrix for predictions, rather than a data frame.
Note
Argument lists and detailed information are available by clicking the
link on the specific method you are currently working with.
For example, if you have a linear model object
testdata.lm <- lm(y ~ x1 + x2, data=testdata)
and you wish to use predict(), then click the link for
predict.lm.
See Also
Examples
# extract the fitted linear predictor from a glm object
glmob <- glm(Kyphosis ~ poly(Age, 2) + (Number > 5)*Start,
family=binomial, data=Sdatasets::kyphosis)
predict(glmob)
newdata <- Sdatasets::kyphosis[11:20,]
predict(glmob, newdata, type = "terms")