Building Data Mining Tools Using JDM - Java Data Mining: Strategy, Standard, and Practice - page 331

interface allows reusing a previously built model to compute quanti-
tative quality metrics on a dataset specified by the user.
Whenever possible, we highlight good JDM programming prac-
tices, such as use of the verify method to minimize the chance of
exceptions after the tasks have been scheduled.

13.1.1 Architecture of the Demonstration Interfaces

The three graphical user interfaces, referred to as consoles hereafter,
need five graphical windows, as shown in Figure 13-1.
The demonstration interface source code can be found under http://
www.kxen.com/products/analytic_framework/kjdm.php . This source code is
organized into the following files:

• JDMConsole.java contains the common methods defined
for all three user interfaces. The class JDMConsole can serve
as the basis for any Java user interface looking like a console
(see following text).
• JDMAdminConsole.java contains the class defining the user
interface to browse the objects persisted in the MOR.
• JDMBuildConsole.java contains the class defining the user
interface to build models of all data mining functions sup-
ported by the DME.
• JDMTestConsole.java contains the class defining the user
interface to apply a model previously built on new datasets.

These files are complemented by the following utility files:

• DescriptiveStatisticDisplayer.java contains the class defin-
ing the user interface to visualize descriptive statistics.
• TestMetricsDisplayer.java contains the class defining the
user interface to display quantitative metrics about classifi-
cation and regression models.

Finally, this file deals with exception handling:

• UnexpectedJDMException.java is another Java RuntimeEx-
ception used to encapsulate JDM exceptions. In Java, a Run-
timeException does not have to be explicitly declared and this
feature can be beneficial, as discussed in the next section.