/* (non-Javadoc)
* @see categorizer.core.Categorizer#load(common.DataContext)
*/
public void load(DataContext categorizerDataContext) throws Exception {
super.load(categorizerDataContext);
DataContext catSpecific = categorizerDataContext.getNode(categorizerSpecificTag);
priors = catSpecific.getNode(priorsTag);
Vector tempVector = catSpecific.getElements2(classIndexTag);
if (tempVector != null && tempVector.size() > 0) {
for (int i = 0; i < dataSet.getDataHeaders().length; i++) {
if (dataSet.getDataHeaders()[i].getLabel().equals((String) tempVector.get(0))) {
classIndex = i;
break;
}
}
}
valid = true;
}
/**
* Builds the naive bayes classifier by calculating prior and conditional probabilities.
*
* @see categorizer.aiCategorizer.core.AICategorizer#buildCategorizer()
*/
@Override
public ConfusionMatrix buildCategorizer() throws Exception {
classIndex = dataSet.getClassIndex();
if (classIndex == -1) {
for (int i = dataSet.getDataHeaders().length - 1; i >= 0; i--) {
if (dataSet.getDataHeaders()[i].isNominal()) {
classIndex = i;
break;
}
}
}
classes = dataSet.getDataHeaders()[classIndex].getAvailableValue();
if (!dataSet.getDataHeaders()[classIndex].isNominal()) throw new ClassNotNominalException();
for (int i = 0; i < classes.length; i++) {
priors.add(new NodePair(classes[i], String.valueOf(0)));
System.out.println(classes[i]);
}
System.out.println("#######################");
for (int i = 0; i < dataSet.getDataItems().length; i++) {
if (!dataSet.getDataItems()[i].getDataFields()[classIndex].getStringValue().equals("?")) {
Vector tempVector =
priors.getElements2(
dataSet.getDataItems()[i].getDataFields()[classIndex].getStringValue());
if (tempVector != null && tempVector.size() >= 0) {
int count = Integer.parseInt((String) tempVector.get(0));
count++;
priors.remove(dataSet.getDataItems()[i].getDataFields()[classIndex].getStringValue());
priors.add(
new NodePair(
dataSet.getDataItems()[i].getDataFields()[classIndex].getStringValue(),
String.valueOf(count)));
}
}
//
// System.out.println(dataSet.getDataItems()[i].getDataFields()[dataSet.getDataHeaders().length - 1].getStringValue());
}
for (int i = 0; i < classes.length; i++) {
int count = 0;
Vector tempVector = priors.getElements2(classes[i]);
if (tempVector != null && tempVector.size() >= 0)
count = Integer.parseInt((String) tempVector.get(0)) + 1;
priors.remove(classes[i]);
priors.add(
new NodePair(
classes[i],
String.valueOf(((double) count) / (dataSet.getDataItems().length + classes.length))));
}
DistributionFinder distFinder = new DistributionFinder();
distFinder.findDistributions(dataSet, classes);
// return null;
valid = true;
return super.validate(testSet);
}
