private void computeValueHistograms(
RDFDataSource source, RDFDataDescriptor desc, RbcAttribute targetAttribute)
throws RDFDatabaseException {
int numOfClassLabels = targetAttribute.getDomainSize();
int numOfAttributeValues = mAttribute.getDomainSize();
if (mValueHistograms == null) {
mValueHistograms = CollectionUtil.makeList();
for (int j = 0; j < numOfClassLabels; j++) {
Map<URI, Double> histogram = CollectionUtil.makeMap();
mValueHistograms.add(histogram);
}
}
for (int j = 0; j < numOfClassLabels; j++) {
Map<URI, Double> valueHistogram = mValueHistograms.get(j);
for (int k = 0; k < numOfAttributeValues; k++) {
URI key = mCut.get().get(k);
if (valueHistogram.containsKey(key)) continue;
SuffStatQueryParameter queryParam =
new SuffStatQueryParameter(desc.getTargetType(), targetAttribute, j, mAttribute, k);
ISufficentStatistic tempSuffStat = source.getBernoulliSufficientStatistic(queryParam);
double valueCount = tempSuffStat.getValue().intValue();
valueHistogram.put(key, valueCount);
// System.out.println(queryParam);
// System.out.println(tempSuffStat.getValue());
}
}
}
public void estimateParameters(RDFDataSource source, RDFDataDescriptor desc)
throws RDFDatabaseException {
// Explicitly ask for class count since every attribute may be HISTOGRAM
RbcAttribute targetAttribute = desc.getTargetAttribute();
int numOfClassLabels = targetAttribute.getDomainSize();
double[] classCounts = new double[numOfClassLabels];
for (int j = 0; j < numOfClassLabels; j++) {
ISufficentStatistic tempSuffStat =
source.getMultinomialSufficientStatistic(
new SuffStatQueryParameter(desc.getTargetType(), targetAttribute, j));
classCounts[j] = tempSuffStat.getValue().intValue();
}
mClassHistogram = new Histogram(classCounts);
mNumInstances = (int) mClassHistogram.sum();
}