You just built and tested four classification models; now identify the best model.
Right-click ClassBuild
, the classification node. Select Compare Test Results.
The model test viewer launches. The Performance tab compares the results of the four classification models.
There are four measures that describe model test:
For information about these measures, see Classification Test Viewer.
The NB (Naive Bayes) model has high Predictive Confidence. The Naive Bayes algorithm affords fast, highly scalable model building and scoring. For the purposes of this exercise, NB is the best model.
Close the tab containing the viewer.
Select ClassBuild
, the classification node. Click the Models tab in the ClassBuild
Property Inspector.
In the Output column, click the green arrow
for all of the models except for the Naive Bayes (NB) model. The green arrow changes to
; models marked this way are ignored, that is,
not passed to subsequent nodes.
The
Next step is to save the details of the selected model.
Copyright © 2011, 2012, Oracle. All rights reserved.