Java Reference
In-Depth Information
101. //Display cluster details
102. Cluster c = (Cluster)clusterIterator.next();
103. System.out.println("Cluster: " + c.getClusterId());
104. System.out.println("Number of cases: " + c.getCaseCount());
105. //Display age attribute details of the cluster
106. Double centroidValue = c. getCentroidCoordinate("age");
107. System.out.println("CentroidCoordinate for Age attribute: "
108. + centroidValue.toString());
109. AttributeStatisticsSet clusterAttrStatitics = c.getStatistics();
110. UnivariateStatitics ageAttrStatistics =
111. clusterAttrStatitics.getStatistics("age");
112. if(ageAttrStatistics != null) {
113. System.out.println("Age attribute values distribution:");
114. Object[] ageValues = ageAttrStatistics.getValues();
115. long[] frequencies = ageAttrStatistics.getFrequencies();
116. for(int i=0; i< ageValues.length; i++)
117. System.out.println("Age:" + ageValues[I].toString() + " -> Frequency " +
118. frequencies[i]);
119. }}}}
9.9
Summary
JDM provides a comprehensive API to build data mining solutions
and tools. JDM provides an API to build, apply, test, and describe
models. Vendors can easily extend the standard API to add more
implementation-specific non-JDM standard functions, algorithms,
and settings. The Connection interface provides methods for inter-
acting with the DME, such as saving and exploring mining objects,
and executing mining tasks. Data specification interfaces provide
various types of data description capabilities and function and algo-
rithm level settings specification. JDM supports specifying minimal
settings for mining operations via simple object representations and
DME defaults. JDM also supports highly customized settings with
options for detailed DME control. For descriptive models, JDM pro-
vides the function and algorithm-specific model details that appli-
cations can retrieve and present to the user. For supervised and
clustering models, the apply task can produce apply outputs for
multi-record or single-record data, supporting batch and real-time
applications, respectively. Test metrics can be computed to evaluate
the quality of supervised models.
 
Search WWH ::




Custom Search