public Instances transformInstances(
Instances inst, Instances output, Map<FV, Collection<FV>> map) {
Set<FV> set = map.keySet();
Double[] substitution = calculateAverage(inst);
// Prepare the list
// First level indicate which attribute the FVS resides in
List<List<Value>> list = new ArrayList<>();
for (int i = 0; i < inst.numAttributes(); i++) {
list.add(new ArrayList<Value>());
}
// Build the data structure
for (FV fv : set) {
list.get(fv.getFeature()).add(new Value(fv.getValue().toString()));
}
for (int i = 0; i < inst.numInstances(); i++) {
Instance instance = getFVSFilteredInstance(output, inst.instance(i), list, substitution);
output.add(instance);
}
return output;
}
public Map<FV, Collection<FV>> extractValuesFromData(Instances inst) {
Multimap<FV, FV> fv_list = ArrayListMultimap.create();
// Instances outFormat = getOutputFormat();
for (int i = 0; i < inst.numInstances(); i++) {
Instance ins = inst.instance(i);
// Skip the class label
for (int x = 0; x < ins.numAttributes() - 1; x++) {
Object value = null;
try {
value = ins.stringValue(x);
} catch (Exception e) {
value = ins.value(x);
}
FV fv = new FV(x, value, ins.classValue());
fv.setNumLabels(inst.numClasses());
if (!fv_list.put(fv, fv)) {
System.err.println("Couldn't put duplicates: " + fv);
}
}
}
Map<FV, Collection<FV>> original_map = fv_list.asMap();
return original_map;
}
public List<Double> generateEntropy(Map<FV, Collection<FV>> fv_list, int numInstances) {
List<Double> entropies = new ArrayList();
Iterator<Entry<FV, Collection<FV>>> iterator = fv_list.entrySet().iterator();
while (iterator.hasNext()) {
Entry<FV, Collection<FV>> next = iterator.next();
FV key = next.getKey();
key.setFrequency((double) next.getValue().size() / numInstances);
double[] counter = new double[key.getNumLabels()];
for (FV fv : next.getValue()) {
int idx = (int) fv.getLabel();
counter[idx]++;
}
key.setEntropy(calculateEntropy(counter, next.getValue().size()));
entropies.add(key.getEntropy());
}
return entropies;
}