|
Options for the algorithm and optional parameters should be collected in a list object. This allows us to set or to modify those options which are necessary. Almost all functions in the glm library allow options. It is possible to give the same list of options to different functions. For example,
opt=glmopt("miter",20,"name","MyDisplay")will set the maximal number of iteration to 20 and the name of the output display to MyDisplay. Now, one can call first
l=glmest("bilo",x,y,opt) glmout("bilo",x,y,l.b,l.bv,l.stat,opt)Both
Principally, it is possible to define the list of optional parameters with
the XploRe command list.
However, it is recommended to use the
glmopt
tool
to set the options. The first call of
glmopt
will create a list of options. To append a further component
to opt, we have to repeat the name
opt as the last argument of
glmopt. For example,
opt=glmopt("miter",20,"title","MyDataset") opt=glmopt("name","MyDisplay",opt)creates the list opt with a component miter containing the value 20, a component title containing the string "MyDataset" and a component name containing the string "MyDisplay". The resulting list has three components as one can check with names(opt).
The next sections will explain which options can be used in GLM estimation.
Prior weights and offsets can always be given as an optional parameter. The corresponding components of the list of optional parameters are
There is a number of control parameters which modify the used algorithm:
The following parameters switch on/off information during the computation.
The model selection functions
glmselect,
glmforward
and
glmbackward
also provide two extra parameters:
Functions which provide graphical output (
glmout,
glmplot,
doglm) accept special options to
change output.
![]() |
MD*TECH Method and Data Technologies |
http://www.mdtech.de mdtech@mdtech.de |