Building Data Mining Tools Using JDM - Java Data Mining: Strategy, Standard, and Practice - page 373

the associated dialog can be filled directly with the five scalar values,
as shown here:

1. public TestMetricsDisplayer(RegressionTestMetrics iTestMetrics) {
2. setLayout(new GridLayout(5, 2));

3. add(new JLabel("MeanAbsoluteError: "));

4. add(new JLabel(iTestMetrics.getMeanAbsoluteError().toString()));

5. add(new JLabel("MeanActualValue: "));

6. add(new JLabel(iTestMetrics.getMeanActualValue().toString()));

7. add(new JLabel("MeanPredictedValue: "));

8. add(new JLabel(iTestMetrics.getMeanPredictedValue().toString()));

9. add(new JLabel("RMSError: "));

10. add(new JLabel(iTestMetrics.getRMSError().toString()));

11. add(new JLabel("RSquared: "));

12. add(new JLabel(iTestMetrics.getRSquared().toString()));

13.}

This concludes the explanation of the source code of the JDMTest-
Console .

13.5 Summary

This chapter presented the code for some graphical interfaces to
illustrate how data mining tools can be developed using JDM. The
source code can be freely downloaded and modified. The chapter
developed three user interfaces: the first deals with model and task
management, the second with data exploration and model building,
and the third with quantitative model validation. They should be
complemented later with tools to allow easy model apply and
export/import of models to cover the deployment features currently
offered by JDM. These are left as exercises for the reader.
Some extensions could be added to deal with data preparation that
will be supported by JDM 2.0; this will allow a complete set of func-
tions currently available in commercial and free data mining tools.
It is important to note that these data mining tools should work on
any implementation compliant with JDM. Moreover, these tools could
be used as seeds for more sophisticated data mining software, operat-
ing across different JDM implementations. These tools provide a user-
friendly way to compare DME implementation behaviors such as:

• The behavior of DME implementations in the presence of
errors (for example, user-friendly error messages, or type of
exceptions raised)