/**
* Shows histogram for selected binned attributes
*
* @param binningType type of discretization performed i.e., quantile, equal width or custom
* @param xformResult name of the result discretized view
* @param attributes names of attributes for which histogram is displayed
*/
public static void displayDiscretizationResults(
OraNumericalBinningType binningType, String xformResult, String[] attributes) {
System.out.println("\nShowing results of the discretization transformation");
System.out.println("\tType of discretization: " + binningType.name());
java.sql.Connection dbConn = ((OraConnection) m_dmeConn).getDatabaseConnection();
for (int i = 0; i < attributes.length; i++) {
String sqlQuery =
MessageFormat.format(
"SELECT {0} BIN_NUMBER, COUNT(*) FREQUENCY FROM ({1}) GROUP BY {0} "
+ "ORDER BY FREQUENCY DESC,BIN_NUMBER ASC",
new String[] {"\"" + attributes[i] + "\"", xformResult});
Statement stmt = null;
ResultSet rs = null;
System.out.println("\tHistogram for:" + attributes[i]);
try {
stmt = dbConn.createStatement();
rs = stmt.executeQuery(sqlQuery);
while (rs.next()) {
String binValue = rs.getString("BIN_NUMBER");
int freq = rs.getInt("FREQUENCY");
System.out.println("\t\t" + binValue + " " + freq);
}
} catch (Exception e) {
System.out.println(e);
} finally {
try {
rs.close();
stmt.close();
} catch (Exception e) {
}
}
}
}
/**
* Illustrates how to perform data discretization.
*
* @param resultXformName name of the result discretized view
* @param binningType type of discretization to perform i.e., quantile, equal width or custom
* @throws JDMException if discretization failed
*/
public static void binData(String resultXformName, OraNumericalBinningType binningType)
throws JDMException {
// Schema where the original data and resulting transformations reside
String schema = (m_dmeConn.getConnectionSpec().getName()).toUpperCase();
// Create discretization transformation instance
OraBinningTransform obt = m_xformFactory.createBinningTransform();
obt.setTransformInputData(schema + "." + "MINING_DATA_BUILD_V");
obt.setTransformOutputData(schema + "." + resultXformName);
// Specify the number of numeric bins
obt.setNumberOfBinsForNumerical(10);
// Specify the number of categoric bins
obt.setNumberOfBinsForCategorical(8);
// Specify the list of excluded attributes
String[] excludedList = new String[] {"CUST_ID", "CUST_GENDER"};
obt.setExcludeColumnList(excludedList);
// Specify the type of numeric binning: equal-width or quantile
// ( default is quantile )
obt.setNumericalBinningType(binningType);
// Specify the type of categorical binning as Top-N: by default it is none
obt.setCategoricalBinningType(OraCategoricalBinningType.top_n);
ArrayList xformList = new ArrayList();
xformList.add(obt);
// Create a transformation sequence object
OraTransformationSequence xformSeq =
m_xformFactory.createTransformationSequence(
schema + "." + "MINING_DATA_BUILD_V", // name of the input data set
xformList, // List of transformations. In this case only one type of transformation
// i.e., supervised binning
schema + "." + resultXformName // name of the transformation result
);
String xformSeqName = "bin_" + binningType.name() + "_xfSeq";
m_dmeConn.saveObject(xformSeqName, xformSeq, true);
OraTransformationTask xformTask = m_xformTaskFactory.create(xformSeqName, false);
executeTask(xformTask, "xFormBin_jdm");
displayDiscretizationResults(
binningType,
schema + "." + resultXformName,
new String[] {"CUST_INCOME_LEVEL", "OCCUPATION"});
}