Ignore:
Timestamp:
09/24/15 10:59:05 (9 years ago)
Author:
sherbold
Message:
  • formatted code and added copyrights
File:
1 edited

Legend:

Unmodified
Added
Removed

  • trunk/CrossPare/src/de/ugoe/cs/cpdp/dataprocessing/BiasedWeights.java

    r2 r41  
     1// Copyright 2015 Georg-August-Universität Göttingen, Germany 
     2// 
     3//   Licensed under the Apache License, Version 2.0 (the "License"); 
     4//   you may not use this file except in compliance with the License. 
     5//   You may obtain a copy of the License at 
     6// 
     7//       http://www.apache.org/licenses/LICENSE-2.0 
     8// 
     9//   Unless required by applicable law or agreed to in writing, software 
     10//   distributed under the License is distributed on an "AS IS" BASIS, 
     11//   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 
     12//   See the License for the specific language governing permissions and 
     13//   limitations under the License. 
     14 
    115package de.ugoe.cs.cpdp.dataprocessing; 
    216 
    … …  
    721 
    822/** 
    9  * Sets the bias of the weights of the training data. By using a bias of 0.5 (default value) the total weight of the positive instances (i.e. 
    10  * fault-prone) is equal to the total weight of the negative instances (i.e. non-fault-prone). Otherwise the weights between the two will be  
    11  * distributed according to the bias, where <0.5 means in favor of the negative instances and >0.5 in favor of the positive instances.  
    12  * equal to the total weight of the test  
     23 * Sets the bias of the weights of the training data. By using a bias of 0.5 (default value) the 
     24 * total weight of the positive instances (i.e. fault-prone) is equal to the total weight of the 
     25 * negative instances (i.e. non-fault-prone). Otherwise the weights between the two will be 
     26 * distributed according to the bias, where <0.5 means in favor of the negative instances and 
     27 * >0.5 in favor of the positive instances. equal to the total weight of the test 
     28 *  
    1329 * @author Steffen Herbold 
    1430 */ 
    1531public class BiasedWeights implements IProcessesingStrategy, ISetWiseProcessingStrategy { 
    1632 
    17         /** 
    18          * bias used for the weighting 
    19          */ 
    20         private double bias = 0.5; 
    21          
    22          
    23         /** 
    24          * Sets the bias to be used for weighting. 
    25          * @param parameters string with the bias 
    26          */ 
    27         @Override 
    28         public void setParameter(String parameters) { 
    29                 bias = Double.parseDouble(parameters); 
    30         } 
     33    /** 
     34     * bias used for the weighting 
     35     */ 
     36    private double bias = 0.5; 
    3137 
    32         /** 
    33          * @see de.ugoe.cs.cpdp.dataprocessing.ProcessesingStrategy#apply(weka.core.Instances, weka.core.Instances) 
    34          */ 
    35         @Override 
    36         public void apply(Instances testdata, Instances traindata) { 
    37                 //setBiasedWeights(testdata); 
    38                 setBiasedWeights(traindata); 
    39         } 
    40          
    41         /** 
    42          * @see de.ugoe.cs.cpdp.dataprocessing.SetWiseProcessingStrategy#apply(weka.core.Instances, org.apache.commons.collections4.list.SetUniqueList) 
    43          */ 
    44         @Override 
    45         public void apply(Instances testdata, SetUniqueList<Instances> traindataSet) { 
    46                 for( Instances traindata : traindataSet ) { 
    47                         setBiasedWeights(traindata); 
    48                 } 
    49         } 
    50          
    51         /** 
    52          * Helper method that sets the weights for a given data set. 
    53          * @param data data set whose weights are set 
    54          */ 
    55         private void setBiasedWeights(Instances data) { 
    56                 final int classIndex = data.classIndex(); 
    57                  
    58                 final int[] counts = data.attributeStats(classIndex).nominalCounts; 
    59                  
    60                 final double weightNegatives = ((1-bias)*data.numInstances()) / counts[0]; 
    61                 final double weightPositives = (bias*data.numInstances()) / counts[1]; 
    62                  
    63                  
    64                 for( int i=0 ; i<data.numInstances() ; i++ ) { 
    65                         Instance instance = data.instance(i); 
    66                         if( instance.value(classIndex)==0 ) { 
    67                                 instance.setWeight(weightNegatives); 
    68                         } 
    69                         if( instance.value(classIndex)==1 ) { 
    70                                 instance.setWeight(weightPositives); 
    71                         } 
    72                 } 
    73         } 
     38    /** 
     39     * Sets the bias to be used for weighting. 
     40     *  
     41     * @param parameters 
     42     *            string with the bias 
     43     */ 
     44    @Override 
     45    public void setParameter(String parameters) { 
     46        bias = Double.parseDouble(parameters); 
     47    } 
    7448 
    75          
     49    /** 
     50     * @see de.ugoe.cs.cpdp.dataprocessing.ProcessesingStrategy#apply(weka.core.Instances, 
     51     *      weka.core.Instances) 
     52     */ 
     53    @Override 
     54    public void apply(Instances testdata, Instances traindata) { 
     55        // setBiasedWeights(testdata); 
     56        setBiasedWeights(traindata); 
     57    } 
     58 
     59    /** 
     60     * @see de.ugoe.cs.cpdp.dataprocessing.SetWiseProcessingStrategy#apply(weka.core.Instances, 
     61     *      org.apache.commons.collections4.list.SetUniqueList) 
     62     */ 
     63    @Override 
     64    public void apply(Instances testdata, SetUniqueList<Instances> traindataSet) { 
     65        for (Instances traindata : traindataSet) { 
     66            setBiasedWeights(traindata); 
     67        } 
     68    } 
     69 
     70    /** 
     71     * Helper method that sets the weights for a given data set. 
     72     *  
     73     * @param data 
     74     *            data set whose weights are set 
     75     */ 
     76    private void setBiasedWeights(Instances data) { 
     77        final int classIndex = data.classIndex(); 
     78 
     79        final int[] counts = data.attributeStats(classIndex).nominalCounts; 
     80 
     81        final double weightNegatives = ((1 - bias) * data.numInstances()) / counts[0]; 
     82        final double weightPositives = (bias * data.numInstances()) / counts[1]; 
     83 
     84        for (int i = 0; i < data.numInstances(); i++) { 
     85            Instance instance = data.instance(i); 
     86            if (instance.value(classIndex) == 0) { 
     87                instance.setWeight(weightNegatives); 
     88            } 
     89            if (instance.value(classIndex) == 1) { 
     90                instance.setWeight(weightPositives); 
     91            } 
     92        } 
     93    } 
    7694 
    7795} 
Note: See TracChangeset for help on using the changeset viewer.