These functions are used by the Generalized Linear Models,
Logistic Regression and Log-linear Regression dialogs.
menuGam
calls
tabSummary.gam,
tabPlot.gam and
tabPredict.gam
if summary, plotting and prediction results are requested.
a formula object, with the response on the left of a `~' operator,
and the terms, separated by
+ operators, on the right.
gamobj
an object that inherits from class
gam.
OPTIONAL ARGUMENTS:
family
the name of a family object which is
a family object -- a list of functions and expressions
for defining the
link and
variance functions, initialization
and iterative weights. Families supported are
gaussian,
binomial,
poisson,
Gamma,
inverse.gaussian and
quasi.
link
if the
family object takes a link argument it should be given here.
variance
if the
family object takes a variance argument it should be given here.
data
an optional data frame in which to interpret the variables occurring in the
formula.
weights
the optional weights for the fitting criterion.
subset
an expression saying which subset of the rows of the data
should be used in the fit.
This can be a logical vector (which is replicated to have length equal
to the number of observations), or a numeric vector indicating which
observation numbers are to be included, or a character vector of the
row names to be included.
All observations are included by default.
na.omit.p
if
TRUE, then any observation with missing values are removed
from the analysis.
If
FALSE and there are missing values then the function will exit
with a message that missing values are not allowed.
If
na.omit.p is
TRUE then
na.action is set to
na.omit in the
call to
gam.
If
na.omit.p is
FALSE then
na.action is set to
na.fail in the
call to
gam.
the convergence threshold for backfitting iterations.
print.short.p
if
TRUE, a short summary of the generalized linear model is printed.
This output is from the function
print.gam.
print.long.p
if
TRUE, a long summary of the generalized linear model is printed.
This output is from the function
summary.gam.
save.name
a character string for the name of the data frame to save the
fit and residuals in.
If data frame with this name already exists in database 1 and it has the
appropriate number of rows then the saved values will be appended
to the data frame.
If the object already exist in database 1 and it is not a data frame
or it does not have the appropriate number of rows then a new name
is created by appending a number to
save.name and the results are
saved in the data frame with the new name.
save.fit.p
if
TRUE, the fitted values from the regression are saved in the
data frame
save.name.
save.resid.working.p
if
TRUE, the working residuals from the model are saved in the
data frame
save.name.
These are the residuals from the final additive fit.
save.resid.pearson.p
if
TRUE, the pearson residuals are saved in the data frame
save.name.
These are standardized residuals on the scale of the response.
save.resid.deviance.p
if
TRUE, the deviance residuals are saved in the data frame
save.name.
The sum of squares of these add up to the deviance.
save.resid.response.p
if
TRUE, the response residuals are saved in the data frame
save.name.
These are the response minus the fitted value.
plotResidVsFit.p
if
TRUE, a plot of the deviance residuals versus the fitted values is created.
plotSqrtAbsResid.p
if
TRUE, a plot of the absolute value of the square root of the
deviance residuals versus the fitted values is created.
This plot is useful for checking for the constant variance assumption
of the model.
plotResponseVsFit.p
if
TRUE, a plot of the response versus the fitted values is created.
plotQQ.p
if
TRUE, a Normal quantile-quantile plot of the pearson residuals is created.
smooths.p
if
TRUE a smooth curve, computed with
loess.smooth
is displayed on the Residuals vs Fit,
Sqrt Abs Residuals vs Fit and Response vs Fit plots.
rugplot.p
if
TRUE, a rugplot is displayed on the Residuals vs Fit,
Sqrt Abs Residuals vs Fit and Response vs Fit plots.
A rugplot is a sequence of vertical bars along the x-axis that mark the
"observed" x values.
id.n
the number of extreme points that will be identified on the
Residuals vs Fit, Sqrt Abs Residuals vs Fit, and Residual's Normal QQ plots.
The row names from the model's data frame
will be used to identify the points.
plotPartialResid.p
if
TRUE, partial residual plots for all the terms in the model
will be created.
plotPartialFit.p
if
TRUE, the partial fit for the term is also displayed on the
partial residual plots.
rugplotPartialResid.p
if
TRUE, a rugplot is displayed on the partial residual plots.
A rugplot is a sequence of vertical bars along the x-axis that mark the
"observed" x values.
scalePartialResid.p
if
TRUE, all the partial residual plots will have the same vertical units.
This is essential for comparing the importance of fitted terms
in additive models.
newdata
a data frame to use for computing predictions.
It must contain the same names as the terms in the right side of the
formula for the model.
If missing, the predictions for the original data are computed.
predobj.name
a character string for the name of the data frame to save the
predictions, standard errors and confidence intervals in.
If data frame with this name already exists in database 1 and it has the
appropriate number of rows then the values will be appended
to the data frame.
If the object already exist in database 1 and it is not a data frame
or it does not have the appropriate number of rows then a new name
is created by appending a number to
predobj.name and the values are
saved in data frame with the new name.
predict.type
a character string denoting the type of prediction to be saved.
The posssible values are:
"link" for predictions are on the additive predictor (link) scale,
"response" for predictions are on the response scale using the inverse
link function
and
"terms" for a matrix of predictions, one for each term in the model.
predict.p
if
TRUE, the predicted values are saved in
predobj.name.
se.p
if
TRUE, the pointwise standard errors for the predictions will be
stored in
predobj.name.
VALUE:
an object of class
gam. See the
gam.object help file for details.
SIDE EFFECTS:
Plots will be drawn if requested.
The objects
save.name and
predobj.name will be created or appended
to if fitted values, residuals or predictions are saved.