See predict.pffr for alternative options to extract estimated
values from a pffr object.
"Fitted values" here refers to the estimated additive predictor values,
these will not be on the scale of the response for models with link functions.
# S3 method for pffr
residuals(object, reformat = TRUE, ...)# S3 method for pffr
fitted(object, reformat = TRUE, which = c("mean", "scale", "both"), ...)
A matrix or ydata-like data.frame or a vector of
residuals / fitted values (see reformat-argument). For
fitted.pffr with family = "gaulss" and which = "both",
returns a list with mean and scale components.
a fitted pffr-object
logical, defaults to TRUE. Should residuals/fitted values be returned in
n x yindex matrix form (regular grid data) or, respectively, in the
shape of the originally supplied ydata argument (sparse/irregular
data), or, if FALSE, simply as a long vector as returned by
resid.gam() or fitted.gam()?
other arguments, passed to residuals.gam.
For fitted.pffr with family = "gaulss" only:
which fitted values to return. One of "mean" (default, returns
predicted means), "scale" (returns predicted log-standard
deviations), or "both" (returns list with both components).
Fabian Scheipl
For family = "gaulss" (Gaussian location-scale models), the fitted
values matrix has two columns: means and log-standard deviations. Use the
which argument in fitted.pffr to control which values are
returned.