Coverage Report

Coverage Report - weka.classifiers.bayes.net.estimate.DiscreteEstimatorBayes

Classes in this File

Line Coverage

Branch Coverage

Complexity

DiscreteEstimatorBayes

0/68

0/36

3.667

 /*
  *   This program is free software: you can redistribute it and/or modify
  *   it under the terms of the GNU General Public License as published by
  *   the Free Software Foundation, either version 3 of the License, or
  *   (at your option) any later version.
  *
  *   This program is distributed in the hope that it will be useful,
  *   but WITHOUT ANY WARRANTY; without even the implied warranty of
  *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  *   GNU General Public License for more details.
  *
  *   You should have received a copy of the GNU General Public License
  *   along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
 /*
  * DiscreteEstimatorBayes.java
  * Adapted from DiscreteEstimator.java
  * Copyright (C) 2012 University of Waikato, Hamilton, New Zealand
  * 
  */
 package weka.classifiers.bayes.net.estimate;
 
 import weka.classifiers.bayes.net.search.local.Scoreable;
 import weka.core.RevisionUtils;
 import weka.core.Statistics;
 import weka.core.Utils;
 import weka.estimators.DiscreteEstimator;
 import weka.estimators.Estimator;
 
 /**
  * Symbolic probability estimator based on symbol counts and a prior.
  * 
  * @author Remco Bouckaert (rrb@xm.co.nz)
  * @version $Revision: 8034 $
  */
 public class DiscreteEstimatorBayes extends Estimator
   implements Scoreable {
 
   /** for serialization */
   static final long serialVersionUID = 4215400230843212684L;
   
   /**
    * Hold the counts
    */
   protected double[] m_Counts;
 
   /**
    * Hold the sum of counts
    */
   protected double   m_SumOfCounts;
 
   /**
    * Holds number of symbols in distribution
    */
   protected int      m_nSymbols = 0;
 
   /**
    * Holds the prior probability
    */
   protected double   m_fPrior = 0.0;
 
   /**
    * Constructor
    * 
    * @param nSymbols the number of possible symbols (remember to include 0)
    * @param fPrior
    */
   public DiscreteEstimatorBayes(int nSymbols, double fPrior) {
     m_fPrior = fPrior;
     m_nSymbols = nSymbols;
     m_Counts = new double[m_nSymbols];
 
     for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
       m_Counts[iSymbol] = m_fPrior;
     } 
 
     m_SumOfCounts = m_fPrior * (double) m_nSymbols;
   }    // DiscreteEstimatorBayes
 
   /**
    * Add a new data value to the current estimator.
    * 
    * @param data the new data value
    * @param weight the weight assigned to the data value
    */
   public void addValue(double data, double weight) {
     m_Counts[(int) data] += weight;
     m_SumOfCounts += weight;
   } 
 
   /**
    * Get a probability estimate for a value
    * 
    * @param data the value to estimate the probability of
    * @return the estimated probability of the supplied value
    */
   public double getProbability(double data) {
     if (m_SumOfCounts == 0) {
 
       // this can only happen if numSymbols = 0 in constructor
       return 0;
     } 
 
     return (double) m_Counts[(int) data] / m_SumOfCounts;
   } 
 
   /**
    * Get a counts for a value
    * 
    * @param data the value to get the counts for
    * @return the count of the supplied value
    */
   public double getCount(double data) {
     if (m_SumOfCounts == 0) {
       // this can only happen if numSymbols = 0 in constructor
       return 0;
     } 
 
     return m_Counts[(int) data];
   } 
   
   /**
    * Gets the number of symbols this estimator operates with
    * 
    * @return the number of estimator symbols
    */
   public int getNumSymbols() {
     return (m_Counts == null) ? 0 : m_Counts.length;
   } 
 
   /**
    * Gets the log score contribution of this distribution
    * @param nType score type
    * @return the score
    */
   public double logScore(int nType, int nCardinality) {
             double fScore = 0.0;
 
             switch (nType) {
 
             case (Scoreable.BAYES): {
               for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
                 fScore += Statistics.lnGamma(m_Counts[iSymbol]);
               } 
 
               fScore -= Statistics.lnGamma(m_SumOfCounts);
               if (m_fPrior != 0.0) {
                       fScore -= m_nSymbols * Statistics.lnGamma(m_fPrior);
                       fScore += Statistics.lnGamma(m_nSymbols * m_fPrior);
               }
             } 
 
               break;
                   case (Scoreable.BDeu): {
                   for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
                         fScore += Statistics.lnGamma(m_Counts[iSymbol]);
                   } 
 
                   fScore -= Statistics.lnGamma(m_SumOfCounts);
                   //fScore -= m_nSymbols * Statistics.lnGamma(1.0);
                   //fScore += Statistics.lnGamma(m_nSymbols * 1.0);
               fScore -= m_nSymbols * Statistics.lnGamma(1.0/(m_nSymbols * nCardinality));
               fScore += Statistics.lnGamma(1.0/nCardinality);
                 } 
                   break;
 
             case (Scoreable.MDL):
 
             case (Scoreable.AIC):
 
             case (Scoreable.ENTROPY): {
               for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
                 double fP = getProbability(iSymbol);
 
                 fScore += m_Counts[iSymbol] * Math.log(fP);
               } 
             } 
 
               break;
 
             default: {}
             }
 
             return fScore;
           } 
 
   /**
    * Display a representation of this estimator
    * 
    * @return a string representation of the estimator
    */
   public String toString() {
     String result = "Discrete Estimator. Counts = ";
 
     if (m_SumOfCounts > 1) {
       for (int i = 0; i < m_Counts.length; i++) {
         result += " " + Utils.doubleToString(m_Counts[i], 2);
       } 
 
       result += "  (Total = " + Utils.doubleToString(m_SumOfCounts, 2) 
                 + ")\n";
     } else {
       for (int i = 0; i < m_Counts.length; i++) {
         result += " " + m_Counts[i];
       } 
 
       result += "  (Total = " + m_SumOfCounts + ")\n";
     } 
 
     return result;
   } 
   
   /**
    * Returns the revision string.
    * 
    * @return                the revision
    */
   public String getRevision() {
     return RevisionUtils.extract("$Revision: 8034 $");
   }
   
   /**
    * Main method for testing this class.
    * 
    * @param argv should contain a sequence of integers which
    * will be treated as symbolic.
    */
   public static void main(String[] argv) {
     try {
       if (argv.length == 0) {
         System.out.println("Please specify a set of instances.");
 
         return;
       } 
 
       int current = Integer.parseInt(argv[0]);
       int max = current;
 
       for (int i = 1; i < argv.length; i++) {
         current = Integer.parseInt(argv[i]);
 
         if (current > max) {
           max = current;
         } 
       } 
 
       DiscreteEstimator newEst = new DiscreteEstimator(max + 1, true);
 
       for (int i = 0; i < argv.length; i++) {
         current = Integer.parseInt(argv[i]);
 
         System.out.println(newEst);
         System.out.println("Prediction for " + current + " = " 
                            + newEst.getProbability(current));
         newEst.addValue(current, 1);
       } 
     } catch (Exception e) {
       System.out.println(e.getMessage());
     } 
   }    // main
  
 }      // class DiscreteEstimatorBayes

1		/*
2		* This program is free software: you can redistribute it and/or modify
3		* it under the terms of the GNU General Public License as published by
4		* the Free Software Foundation, either version 3 of the License, or
5		* (at your option) any later version.
6		*
7		* This program is distributed in the hope that it will be useful,
8		* but WITHOUT ANY WARRANTY; without even the implied warranty of
9		* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
10		* GNU General Public License for more details.
11		*
12		* You should have received a copy of the GNU General Public License
13		* along with this program. If not, see <http://www.gnu.org/licenses/>.
14		*/
15
16		/*
17		* DiscreteEstimatorBayes.java
18		* Adapted from DiscreteEstimator.java
19		* Copyright (C) 2012 University of Waikato, Hamilton, New Zealand
20		*
21		*/
22		package weka.classifiers.bayes.net.estimate;
23
24		import weka.classifiers.bayes.net.search.local.Scoreable;
25		import weka.core.RevisionUtils;
26		import weka.core.Statistics;
27		import weka.core.Utils;
28		import weka.estimators.DiscreteEstimator;
29		import weka.estimators.Estimator;
30
31		/**
32		* Symbolic probability estimator based on symbol counts and a prior.
33		*
34		* @author Remco Bouckaert (rrb@xm.co.nz)
35		* @version $Revision: 8034 $
36		*/
37		public class DiscreteEstimatorBayes extends Estimator
38		implements Scoreable {
39
40		/** for serialization */
41		static final long serialVersionUID = 4215400230843212684L;
42
43		/**
44		* Hold the counts
45		*/
46		protected double[] m_Counts;
47
48		/**
49		* Hold the sum of counts
50		*/
51		protected double m_SumOfCounts;
52
53		/**
54		* Holds number of symbols in distribution
55		*/
56	0	protected int m_nSymbols = 0;
57
58		/**
59		* Holds the prior probability
60		*/
61	0	protected double m_fPrior = 0.0;
62
63		/**
64		* Constructor
65		*
66		* @param nSymbols the number of possible symbols (remember to include 0)
67		* @param fPrior
68		*/
69	0	public DiscreteEstimatorBayes(int nSymbols, double fPrior) {
70	0	m_fPrior = fPrior;
71	0	m_nSymbols = nSymbols;
72	0	m_Counts = new double[m_nSymbols];
73
74	0	for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
75	0	m_Counts[iSymbol] = m_fPrior;
76		}
77
78	0	m_SumOfCounts = m_fPrior * (double) m_nSymbols;
79	0	} // DiscreteEstimatorBayes
80
81		/**
82		* Add a new data value to the current estimator.
83		*
84		* @param data the new data value
85		* @param weight the weight assigned to the data value
86		*/
87		public void addValue(double data, double weight) {
88	0	m_Counts[(int) data] += weight;
89	0	m_SumOfCounts += weight;
90	0	}
91
92		/**
93		* Get a probability estimate for a value
94		*
95		* @param data the value to estimate the probability of
96		* @return the estimated probability of the supplied value
97		*/
98		public double getProbability(double data) {
99	0	if (m_SumOfCounts == 0) {
100
101		// this can only happen if numSymbols = 0 in constructor
102	0	return 0;
103		}
104
105	0	return (double) m_Counts[(int) data] / m_SumOfCounts;
106		}
107
108		/**
109		* Get a counts for a value
110		*
111		* @param data the value to get the counts for
112		* @return the count of the supplied value
113		*/
114		public double getCount(double data) {
115	0	if (m_SumOfCounts == 0) {
116		// this can only happen if numSymbols = 0 in constructor
117	0	return 0;
118		}
119
120	0	return m_Counts[(int) data];
121		}
122
123		/**
124		* Gets the number of symbols this estimator operates with
125		*
126		* @return the number of estimator symbols
127		*/
128		public int getNumSymbols() {
129	0	return (m_Counts == null) ? 0 : m_Counts.length;
130		}
131
132		/**
133		* Gets the log score contribution of this distribution
134		* @param nType score type
135		* @return the score
136		*/
137		public double logScore(int nType, int nCardinality) {
138	0	double fScore = 0.0;
139
140	0	switch (nType) {
141
142		case (Scoreable.BAYES): {
143	0	for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
144	0	fScore += Statistics.lnGamma(m_Counts[iSymbol]);
145		}
146
147	0	fScore -= Statistics.lnGamma(m_SumOfCounts);
148	0	if (m_fPrior != 0.0) {
149	0	fScore -= m_nSymbols * Statistics.lnGamma(m_fPrior);
150	0	fScore += Statistics.lnGamma(m_nSymbols * m_fPrior);
151		}
152		}
153
154		break;
155		case (Scoreable.BDeu): {
156	0	for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
157	0	fScore += Statistics.lnGamma(m_Counts[iSymbol]);
158		}
159
160	0	fScore -= Statistics.lnGamma(m_SumOfCounts);
161		//fScore -= m_nSymbols * Statistics.lnGamma(1.0);
162		//fScore += Statistics.lnGamma(m_nSymbols * 1.0);
163	0	fScore -= m_nSymbols * Statistics.lnGamma(1.0/(m_nSymbols * nCardinality));
164	0	fScore += Statistics.lnGamma(1.0/nCardinality);
165		}
166	0	break;
167
168		case (Scoreable.MDL):
169
170		case (Scoreable.AIC):
171
172		case (Scoreable.ENTROPY): {
173	0	for (int iSymbol = 0; iSymbol < m_nSymbols; iSymbol++) {
174	0	double fP = getProbability(iSymbol);
175
176	0	fScore += m_Counts[iSymbol] * Math.log(fP);
177		}
178		}
179
180	0	break;
181
182		default: {}
183		}
184
185	0	return fScore;
186		}
187
188		/**
189		* Display a representation of this estimator
190		*
191		* @return a string representation of the estimator
192		*/
193		public String toString() {
194	0	String result = "Discrete Estimator. Counts = ";
195
196	0	if (m_SumOfCounts > 1) {
197	0	for (int i = 0; i < m_Counts.length; i++) {
198	0	result += " " + Utils.doubleToString(m_Counts[i], 2);
199		}
200
201	0	result += " (Total = " + Utils.doubleToString(m_SumOfCounts, 2)
202		+ ")\n";
203		} else {
204	0	for (int i = 0; i < m_Counts.length; i++) {
205	0	result += " " + m_Counts[i];
206		}
207
208	0	result += " (Total = " + m_SumOfCounts + ")\n";
209		}
210
211	0	return result;
212		}
213
214		/**
215		* Returns the revision string.
216		*
217		* @return the revision
218		*/
219		public String getRevision() {
220	0	return RevisionUtils.extract("$Revision: 8034 $");
221		}
222
223		/**
224		* Main method for testing this class.
225		*
226		* @param argv should contain a sequence of integers which
227		* will be treated as symbolic.
228		*/
229		public static void main(String[] argv) {
230		try {
231	0	if (argv.length == 0) {
232	0	System.out.println("Please specify a set of instances.");
233
234	0	return;
235		}
236
237	0	int current = Integer.parseInt(argv[0]);
238	0	int max = current;
239
240	0	for (int i = 1; i < argv.length; i++) {
241	0	current = Integer.parseInt(argv[i]);
242
243	0	if (current > max) {
244	0	max = current;
245		}
246		}
247
248	0	DiscreteEstimator newEst = new DiscreteEstimator(max + 1, true);
249
250	0	for (int i = 0; i < argv.length; i++) {
251	0	current = Integer.parseInt(argv[i]);
252
253	0	System.out.println(newEst);
254	0	System.out.println("Prediction for " + current + " = "
255		+ newEst.getProbability(current));
256	0	newEst.addValue(current, 1);
257		}
258	0	} catch (Exception e) {
259	0	System.out.println(e.getMessage());
260	0	}
261	0	} // main
262
263		} // class DiscreteEstimatorBayes