discretize.java

wekaUT是 university texas austin 开发的基于weka的半指导学习(semi supervised learning)的分类器

   * 1. <br>
   * eg: first-3,5,6-last
   * @exception IllegalArgumentException if an invalid range list is supplied 
   */
  public void setAttributeIndices(String rangeList) {

    m_DiscretizeCols.setRanges(rangeList);
  }

  /**
   * Sets which attributes are to be Discretized (only numeric
   * attributes among the selection will be Discretized).
   *
   * @param attributes an array containing indexes of attributes to Discretize.
   * Since the array will typically come from a program, attributes are indexed
   * from 0.
   * @exception IllegalArgumentException if an invalid set of ranges
   * is supplied 
   */
  public void setAttributeIndicesArray(int [] attributes) {

    setAttributeIndices(Range.indicesToRangeList(attributes));
  }

  /**
   * Gets the cut points for an attribute
   *
   * @param the index (from 0) of the attribute to get the cut points of
   * @return an array containing the cutpoints (or null if the
   * attribute requested isn't being Discretized
   */
  public double [] getCutPoints(int attributeIndex) {

    if (m_CutPoints == null) {
      return null;
    }
    return m_CutPoints[attributeIndex];
  }

  /** Generate the cutpoints for each attribute */
  protected void calculateCutPoints() {

    Instances copy = null;

    m_CutPoints = new double [getInputFormat().numAttributes()] [];
    for(int i = getInputFormat().numAttributes() - 1; i >= 0; i--) {
      if ((m_DiscretizeCols.isInRange(i)) && 
	  (getInputFormat().attribute(i).isNumeric())) {

	// Use copy to preserve order
	if (copy == null) {
	  copy = new Instances(getInputFormat());
	}
	calculateCutPointsByMDL(i, copy);
      }
    }
  }

  /**
   * Set cutpoints for a single attribute using MDL.
   *
   * @param index the index of the attribute to set cutpoints for
   */
  protected void calculateCutPointsByMDL(int index,
					 Instances data) {

    // Sort instances
    data.sort(data.attribute(index));

    // Find first instances that's missing
    int firstMissing = data.numInstances();
    for (int i = 0; i < data.numInstances(); i++) {
      if (data.instance(i).isMissing(index)) {
        firstMissing = i;
        break;
      }
    }
    m_CutPoints[index] = cutPointsForSubset(data, index, 0, firstMissing);
  }

  /** Test using Kononenko's MDL criterion. */
  private boolean KononenkosMDL(double[] priorCounts,
				double[][] bestCounts,
				double numInstances,
				int numCutPoints) {

    double distPrior, instPrior, distAfter = 0, sum, instAfter = 0;
    double before, after;
    int numClassesTotal;

    // Number of classes occuring in the set
    numClassesTotal = 0;
    for (int i = 0; i < priorCounts.length; i++) {
      if (priorCounts[i] > 0) {
	numClassesTotal++;
      }
    }

    // Encode distribution prior to split
    distPrior = SpecialFunctions.log2Binomial(numInstances 
					      + numClassesTotal - 1,
					      numClassesTotal - 1);

    // Encode instances prior to split.
    instPrior = SpecialFunctions.log2Multinomial(numInstances,
						 priorCounts);

    before = instPrior + distPrior;

    // Encode distributions and instances after split.
    for (int i = 0; i < bestCounts.length; i++) {
      sum = Utils.sum(bestCounts[i]);
      distAfter += SpecialFunctions.log2Binomial(sum + numClassesTotal - 1,
						 numClassesTotal - 1);
      instAfter += SpecialFunctions.log2Multinomial(sum,
						    bestCounts[i]);
    }

    // Coding cost after split
    after = Utils.log2(numCutPoints) + distAfter + instAfter;

    // Check if split is to be accepted
    return (Utils.gr(before, after));
  }


  /** Test using Fayyad and Irani's MDL criterion. */
  private boolean FayyadAndIranisMDL(double[] priorCounts,
				     double[][] bestCounts,
				     double numInstances,
				     int numCutPoints) {

    double priorEntropy, entropy, gain; 
    double entropyLeft, entropyRight, delta;
    int numClassesTotal, numClassesRight, numClassesLeft;

    // Compute entropy before split.
    priorEntropy = ContingencyTables.entropy(priorCounts);

    // Compute entropy after split.
    entropy = ContingencyTables.entropyConditionedOnRows(bestCounts);

    // Compute information gain.
    gain = priorEntropy - entropy;

    // Number of classes occuring in the set
    numClassesTotal = 0;
    for (int i = 0; i < priorCounts.length; i++) {
      if (priorCounts[i] > 0) {
	numClassesTotal++;
      }
    }

    // Number of classes occuring in the left subset
    numClassesLeft = 0;
    for (int i = 0; i < bestCounts[0].length; i++) {
      if (bestCounts[0][i] > 0) {
	numClassesLeft++;
      }
    }

    // Number of classes occuring in the right subset
    numClassesRight = 0;
    for (int i = 0; i < bestCounts[1].length; i++) {
      if (bestCounts[1][i] > 0) {
	numClassesRight++;
      }
    }

    // Entropy of the left and the right subsets
    entropyLeft = ContingencyTables.entropy(bestCounts[0]);
    entropyRight = ContingencyTables.entropy(bestCounts[1]);

    // Compute terms for MDL formula
    delta = Utils.log2(Math.pow(3, numClassesTotal) - 2) - 
      (((double) numClassesTotal * priorEntropy) - 
       (numClassesRight * entropyRight) - 
       (numClassesLeft * entropyLeft));

    // Check if split is to be accepted
    return (Utils.gr(gain, (Utils.log2(numCutPoints) + delta) /
		     (double)numInstances));
  }
    

  /** Selects cutpoints for sorted subset. */
  private double[] cutPointsForSubset(Instances instances, int attIndex, 
				      int first, int lastPlusOne) { 

    double[][] counts, bestCounts;
    double[] priorCounts, left, right, cutPoints;
    double currentCutPoint = -Double.MAX_VALUE, bestCutPoint = -1, 
      currentEntropy, bestEntropy, priorEntropy, gain;
    int bestIndex = -1, numInstances = 0, numCutPoints = 0;

    // Compute number of instances in set
    if ((lastPlusOne - first) < 2) {
      return null;
    }

    // Compute class counts.
    counts = new double[2][instances.numClasses()];
    for (int i = first; i < lastPlusOne; i++) {
      numInstances += instances.instance(i).weight();
      counts[1][(int)instances.instance(i).classValue()] +=
	instances.instance(i).weight();
    }

    // Save prior counts
    priorCounts = new double[instances.numClasses()];
    System.arraycopy(counts[1], 0, priorCounts, 0, 
		     instances.numClasses());

    // Entropy of the full set
    priorEntropy = ContingencyTables.entropy(priorCounts);
    bestEntropy = priorEntropy;
    
    // Find best entropy.
    bestCounts = new double[2][instances.numClasses()];
    for (int i = first; i < (lastPlusOne - 1); i++) {
      counts[0][(int)instances.instance(i).classValue()] +=
	instances.instance(i).weight();
      counts[1][(int)instances.instance(i).classValue()] -=
	instances.instance(i).weight();
      if (Utils.sm(instances.instance(i).value(attIndex), 
		   instances.instance(i + 1).value(attIndex))) {
	currentCutPoint = instances.instance(i).value(attIndex); //+ 
	//instances.instance(i + 1).value(attIndex)) / 2.0;
	currentEntropy = ContingencyTables.entropyConditionedOnRows(counts);
	if (Utils.sm(currentEntropy, bestEntropy)) {
	  bestCutPoint = currentCutPoint;
	  bestEntropy = currentEntropy;
	  bestIndex = i;
	  System.arraycopy(counts[0], 0, 
			   bestCounts[0], 0, instances.numClasses());
	  System.arraycopy(counts[1], 0, 
			   bestCounts[1], 0, instances.numClasses()); 
	}
	numCutPoints++;
      }
    }

    // Use worse encoding?
    if (!m_UseBetterEncoding) {
      numCutPoints = (lastPlusOne - first) - 1;
    }

    // Checks if gain is zero
    gain = priorEntropy - bestEntropy;
    if (Utils.eq(gain, 0)) {
      return null;
    }

    // Check if split is to be accepted
    if ((m_UseKononenko && KononenkosMDL(priorCounts, bestCounts,
					 numInstances, numCutPoints)) ||
	(!m_UseKononenko && FayyadAndIranisMDL(priorCounts, bestCounts,
					       numInstances, numCutPoints))) {
      
      // Select split points for the left and right subsets
      left = cutPointsForSubset(instances, attIndex, first, bestIndex + 1);
      right = cutPointsForSubset(instances, attIndex, 
				 bestIndex + 1, lastPlusOne);
      
      // Merge cutpoints and return them
      if ((left == null) && (right) == null) {
	cutPoints = new double[1];
	cutPoints[0] = bestCutPoint;
      } else if (right == null) {
	cutPoints = new double[left.length + 1];
	System.arraycopy(left, 0, cutPoints, 0, left.length);
	cutPoints[left.length] = bestCutPoint;
      } else if (left == null) {
	cutPoints = new double[1 + right.length];
	cutPoints[0] = bestCutPoint;
	System.arraycopy(right, 0, cutPoints, 1, right.length);
      } else {
	cutPoints = new double[left.length + right.length + 1];
	System.arraycopy(left, 0, cutPoints, 0, left.length);
	cutPoints[left.length] = bestCutPoint;
	System.arraycopy(right, 0, cutPoints, left.length + 1, right.length);
      }
      
      return cutPoints;
    } else
      return null;
  }
 
  /**
   * Set the output format. Takes the currently defined cutpoints and 
   * m_InputFormat and calls setOutputFormat(Instances) appropriately.
   */
  protected void setOutputFormat() {

    if (m_CutPoints == null) {
      setOutputFormat(null);
      return;
    }
    FastVector attributes = new FastVector(getInputFormat().numAttributes());
    int classIndex = getInputFormat().classIndex();
    for(int i = 0; i < getInputFormat().numAttributes(); i++) {
      if ((m_DiscretizeCols.isInRange(i)) 
	  && (getInputFormat().attribute(i).isNumeric())) {
	if (!m_MakeBinary) {
	  FastVector attribValues = new FastVector(1);
	  if (m_CutPoints[i] == null) {
	    attribValues.addElement("'All'");
	  } else {
	    for(int j = 0; j <= m_CutPoints[i].length; j++) {
	      if (j == 0) {
		attribValues.addElement("'(-inf-"
			+ Utils.doubleToString(m_CutPoints[i][j], 6) + "]'");
	      } else if (j == m_CutPoints[i].length) {
		attribValues.addElement("'("
			+ Utils.doubleToString(m_CutPoints[i][j - 1], 6) 
					+ "-inf)'");
	      } else {
		attribValues.addElement("'("
			+ Utils.doubleToString(m_CutPoints[i][j - 1], 6) + "-"
			+ Utils.doubleToString(m_CutPoints[i][j], 6) + "]'");
	      }
	    }
	  }
	  attributes.addElement(new Attribute(getInputFormat().
					      attribute(i).name(),
					      attribValues));
	} else {
	  if (m_CutPoints[i] == null) {
	    FastVector attribValues = new FastVector(1);
	    attribValues.addElement("'All'");
	    attributes.addElement(new Attribute(getInputFormat().
						attribute(i).name(),
						attribValues));
	  } else {
	    if (i < getInputFormat().classIndex()) {
	      classIndex += m_CutPoints[i].length - 1;
	    }
	    for(int j = 0; j < m_CutPoints[i].length; j++) {
	      FastVector attribValues = new FastVector(2);
	      attribValues.addElement("'(-inf-"
		      + Utils.doubleToString(m_CutPoints[i][j], 6) + "]'");
	      attribValues.addElement("'("
		      + Utils.doubleToString(m_CutPoints[i][j], 6) + "-inf)'");
	      attributes.addElement(new Attribute(getInputFormat().
						  attribute(i).name(),
						  attribValues));
	    }
	  }
	}
      } else {
	attributes.addElement(getInputFormat().attribute(i).copy());
      }
    }
    Instances outputFormat = 
      new Instances(getInputFormat().relationName(), attributes, 0);
    outputFormat.setClassIndex(classIndex);
    setOutputFormat(outputFormat);
  }

  /**
   * Convert a single instance over. The converted instance is added to 
   * the end of the output queue.
   *
   * @param instance the instance to convert
   */
  protected void convertInstance(Instance instance) {

    int index = 0;
    double [] vals = new double [outputFormatPeek().numAttributes()];
    // Copy and convert the values
    for(int i = 0; i < getInputFormat().numAttributes(); i++) {
      if (m_DiscretizeCols.isInRange(i) && 
	  getInputFormat().attribute(i).isNumeric()) {
	int j;
	double currentVal = instance.value(i);
	if (m_CutPoints[i] == null) {
	  if (instance.isMissing(i)) {
	    vals[index] = Instance.missingValue();
	  } else {
	    vals[index] = 0;
	  }
	  index++;
	} else {
	  if (!m_MakeBinary) {
	    if (instance.isMissing(i)) {
	      vals[index] = Instance.missingValue();
	    } else {
	      for (j = 0; j < m_CutPoints[i].length; j++) {
		if (currentVal <= m_CutPoints[i][j]) {
		  break;
		}
	      }
              vals[index] = j;
	    }
	    index++;
	  } else {
	    for (j = 0; j < m_CutPoints[i].length; j++) {
	      if (instance.isMissing(i)) {
                vals[index] = Instance.missingValue();
	      } else if (currentVal <= m_CutPoints[i][j]) {
                vals[index] = 0;
	      } else {
                vals[index] = 1;
	      }
	      index++;
	    }
	  }   
	}
      } else {
        vals[index] = instance.value(i);
	index++;
      }
    }
    
    Instance inst = null;
    if (instance instanceof SparseInstance) {
      inst = new SparseInstance(instance.weight(), vals);
    } else {
      inst = new Instance(instance.weight(), vals);
    }
    copyStringValues(inst, false, instance.dataset(), getInputStringIndex(),
                     getOutputFormat(), getOutputStringIndex());
    inst.setDataset(getOutputFormat());
    push(inst);
  }

  /**
   * Main method for testing this class.
   *
   * @param argv should contain arguments to the filter: use -h for help
   */
  public static void main(String [] argv) {

    try {
      if (Utils.getFlag('b', argv)) {
 	Filter.batchFilterFile(new Discretize(), argv);
      } else {
	Filter.filterFile(new Discretize(), argv);
      }
    } catch (Exception ex) {
      System.out.println(ex.getMessage());
    }
  }
}