package com.yahoo.labs.samoa.moa.classifiers.core.attributeclassobservers;
/*
* #%L
* SAMOA
* %%
* Copyright (C) 2007 University of Waikato, Hamilton, New Zealand
* %%
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
* #L%
*/
import com.yahoo.labs.samoa.moa.core.Utils;
import java.io.Serializable;
import java.util.ArrayList;
import java.util.List;
import com.yahoo.labs.samoa.moa.classifiers.core.AttributeSplitSuggestion;
import com.yahoo.labs.samoa.moa.classifiers.core.conditionaltests.NumericAttributeBinaryTest;
import com.yahoo.labs.samoa.moa.classifiers.core.splitcriteria.SplitCriterion;
import com.yahoo.labs.samoa.moa.core.DoubleVector;
import com.yahoo.labs.samoa.moa.core.ObjectRepository;
import com.yahoo.labs.samoa.moa.options.AbstractOptionHandler;
import com.github.javacliparser.IntOption;
import com.yahoo.labs.samoa.moa.tasks.TaskMonitor;
/**
* Class for observing the class data distribution for a numeric attribute as in VFML.
* Used in naive Bayes and decision trees to monitor data statistics on leaves.
*
* @author Richard Kirkby (rkirkby@cs.waikato.ac.nz)
* @version $Revision: 7 $
*/
public class VFMLNumericAttributeClassObserver extends AbstractOptionHandler implements NumericAttributeClassObserver {
private static final long serialVersionUID = 1L;
@Override
public void observeAttributeTarget(double attVal, double target) {
throw new UnsupportedOperationException("Not supported yet.");
}
protected class Bin implements Serializable {
private static final long serialVersionUID = 1L;
public double lowerBound, upperBound;
public DoubleVector classWeights = new DoubleVector();
public int boundaryClass;
public double boundaryWeight;
}
protected List<Bin> binList = new ArrayList<>();
public IntOption numBinsOption = new IntOption("numBins", 'n',
"The number of bins.", 10, 1, Integer.MAX_VALUE);
@Override
public void observeAttributeClass(double attVal, int classVal, double weight) {
if (!Utils.isMissingValue(attVal)) {
if (this.binList.size() < 1) {
// create the first bin
Bin newBin = new Bin();
newBin.classWeights.addToValue(classVal, weight);
newBin.boundaryClass = classVal;
newBin.boundaryWeight = weight;
newBin.upperBound = attVal;
newBin.lowerBound = attVal;
this.binList.add(newBin);
} else {
// find bin containing new example with binary search
int index = 0;
boolean found = false;
int min = 0;
int max = this.binList.size() - 1;
while ((min <= max) && !found) {
int i = (min + max) / 2;
Bin bin = this.binList.get(i);
if (((attVal >= bin.lowerBound) && (attVal < bin.upperBound))
|| ((i == this.binList.size() - 1)
&& (attVal >= bin.lowerBound) && (attVal <= bin.upperBound))) {
found = true;
index = i;
} else if (attVal < bin.lowerBound) {
max = i - 1;
} else {
min = i + 1;
}
}
boolean first = false;
boolean last = false;
if (!found) {
// determine if it is before or after the existing range
Bin bin = this.binList.get(0);
if (bin.lowerBound > attVal) {
// go before the first bin
index = 0;
first = true;
} else {
// if we haven't found it yet value must be > last bins
// upperBound
index = this.binList.size() - 1;
last = true;
}
}
Bin bin = this.binList.get(index); // VLIndex(ct->bins, index);
if ((bin.lowerBound == attVal)
|| (this.binList.size() >= this.numBinsOption.getValue())) {// Option.getValue())
// {//1000)
// {
// if this is the exact same boundary and class as the bin
// boundary or we aren't adding new bins any more then
// increment
// boundary counts
bin.classWeights.addToValue(classVal, weight);
if ((bin.boundaryClass == classVal)
&& (bin.lowerBound == attVal)) {
// if it is also the same class then special case it
bin.boundaryWeight += weight;
}
} else {
// create a new bin
Bin newBin = new Bin();
newBin.classWeights.addToValue(classVal, weight);
newBin.boundaryWeight = weight;
newBin.boundaryClass = classVal;
newBin.upperBound = bin.upperBound;
newBin.lowerBound = attVal;
double percent = 0.0;
// estimate initial counts with a linear interpolation
if (!((bin.upperBound - bin.lowerBound == 0) || last || first)) {
percent = 1.0 - ((attVal - bin.lowerBound) / (bin.upperBound - bin.lowerBound));
}
// take out the boundry points, they stay with the old bin
bin.classWeights.addToValue(bin.boundaryClass,
-bin.boundaryWeight);
DoubleVector weightToShift = new DoubleVector(
bin.classWeights);
weightToShift.scaleValues(percent);
newBin.classWeights.addValues(weightToShift);
bin.classWeights.subtractValues(weightToShift);
// put the boundry examples back in
bin.classWeights.addToValue(bin.boundaryClass,
bin.boundaryWeight);
// insert the new bin in the right place
if (last) {
bin.upperBound = attVal;
newBin.upperBound = attVal;
this.binList.add(newBin);
} else if (first) {
newBin.upperBound = bin.lowerBound;
this.binList.add(0, newBin);
} else {
newBin.upperBound = bin.upperBound;
bin.upperBound = attVal;
this.binList.add(index + 1, newBin);
}
}
}
}
}
@Override
public double probabilityOfAttributeValueGivenClass(double attVal,
int classVal) {
// TODO: NaiveBayes broken until implemented
return 0.0;
}
@Override
public AttributeSplitSuggestion getBestEvaluatedSplitSuggestion(
SplitCriterion criterion, double[] preSplitDist, int attIndex,
boolean binaryOnly) {
AttributeSplitSuggestion bestSuggestion = null;
DoubleVector rightDist = new DoubleVector();
for (Bin bin : this.binList) {
rightDist.addValues(bin.classWeights);
}
DoubleVector leftDist = new DoubleVector();
for (Bin bin : this.binList) {
leftDist.addValues(bin.classWeights);
rightDist.subtractValues(bin.classWeights);
double[][] postSplitDists = new double[][]{
leftDist.getArrayCopy(), rightDist.getArrayCopy()};
double merit = criterion.getMeritOfSplit(preSplitDist,
postSplitDists);
if ((bestSuggestion == null) || (merit > bestSuggestion.merit)) {
bestSuggestion = new AttributeSplitSuggestion(
new NumericAttributeBinaryTest(attIndex,
bin.upperBound, false), postSplitDists, merit);
}
}
return bestSuggestion;
}
@Override
public void getDescription(StringBuilder sb, int indent) {
// TODO Auto-generated method stub
}
@Override
protected void prepareForUseImpl(TaskMonitor monitor, ObjectRepository repository) {
// TODO Auto-generated method stub
}
}