Building Data Mining Tools Using JDM - Java Data Mining: Strategy, Standard, and Practice - page 345

13.3.1 General Introduction

The proposed graphical user interface is depicted in Figure 13-3. This
GUI, designed to be simple, will not use LogicalData . It exposes the
data mining functions supported by the vendor implementation, and
allows building models for each supported function. A minimum
number of parameters will be provided for each function. For classifi-
cation and regression, parameters include the name of the target
attribute. For attribute importance, parameters include the optional
target, as well as the number of desired top attributes. For clustering,
parameters include the maximum number of clusters. And for asso-
ciation, parameters include the minimum support, minimum confi-
dence, and maximum rule length for association rules.
This user interface also exposes the URI to reference the build
datasets. Recall that the URI syntax for such datasets is vendor-
specific.
To allow experimentation with vendor implementations, we
have added the ability to overwrite all objects saved in the connec-
tion. When the “overwrite” box is unchecked, an exception should
be thrown if the user specifies an object with a name already in
the MOR.

Figure 13-3 A view of the generic interface to build and save

models.