gamopt
presents a convenient tool to create
the option list opt.
| "x" | discrete predictor variables |
| "loc" | indicator for local const, linear or quadratic estimation |
| "tg" | grid |
| "h" | bandwidth for the directions of interest |
| "g" | bandwidth for the directions not of interest |
| "code" | model code |
| "kern" | name of the kernel function |
| "wx" | weights |
| "off" | offset |
| "shf" | indicator to show iterations |
| "miter" | maximal number of iterations |
| "cnv" | convergence criterion |
| "fscor" | Fisher scoring will be used |
| "pow" | power for
gintestpl |
| "nopic" | show no picture |
| "descript" | add descriptive statistics |
| "pl" | partially linear model |
| "name" | output variable name |
| "xvars" | discrete variable names |
| "tvars" | name of t-variable |
| "yvars" | name of y-variable |
| "title" | output picture title |
| "bv" | covariance matrix for b |
opt = gamopt("code","bipro","miter",10,"nopic",1,opt)
appends the optional parameters code, miter and nopic
containing the values "bipro", 10 and 1 to the already existing option
list opt.
Up to 10 optional parameters may be appended at one call. Rerun the
quantlet to extend your option list.
Finally, check the list with the
names
command or by typing its
name.
list. However, this approach has some
drawbacks. First, all elements of the option list would exist twice: as
global objects as well as list components. Also, name conflicts could
arise, since the components need to have specific names to be
correctly identified by the gam
quantlets. Finally, XploRe supports
list
with identical names for different list components. In this case,
only the first of multiple elements with the same name can be
identified.
Therefore, it is recommended to use the quantlet
gamopt
to set the
options.