Java Reference
In-Depth Information
element is of type Header (line 6) and encapsulates the optional
details. It provides human-readable information about the doc-
ument, such as the application that created it, when it was created,
document description, and copyright information.
The object element is of type NamedObject (line 7) and encapsulates
one of the JDM named objects, such as task and build settings. One
JDM element can encapsulate multiple object elements. There is a
required version attribute (line 5) of the JDM element used to specify
the JDM Schema version.
Listing 10-3 illustrates a JDM document that conforms to the JDM
Schema's elements structure. This document encapsulates a classifi-
cation build settings object using the decision tree algorithm. Note
that the buildSettings element (line 4) uses xsi:type
tings” to indicate that it is a ClassificationSettings
complex type. Sim-
”TreeSettings” to indicate that it is a TreeSettings complex
Listing 10-3
Example JDM document (Classification settings object with decision tree algorithm)
<?xml version="1.0"?>
1. < JDM version="1.1"
xmlns:xsi="" >
2. < header >
<copyright>Copyright (c) 2004, 2005, Oracle. All rights reserved.</copyright>
<applicationName>XYZ Predictive Analytics</applicationName>
<applicationVersion>10.2</applicationVersion >
<description>Provides settings to build customer attrition model</description>
</ header >
3. < object >
4. < buildSettings xsi:type="ClassificationSettings"
miningFunction="classification" targetAttributeName="customer_type">
5. < algorithmSettings xsi:type="TreeSettings"
miningAlgorithm="decisionTree" maxDepth="10" />
Schema Types
JDM Schema types have a one-to-one mapping with the JDM API
classes or interfaces. JDM Schema types use the same name of the asso-
ciated Java classes or interfaces, and maintain the same object inherit-
ance and relationships as in the API whenever possible. In addition,
Search WWH ::

Custom Search