Total members 11892 |It is currently Thu Oct 10, 2024 8:17 am Login / Join Codemiles

Java

C/C++

PHP

C#

HTML

CSS

ASP

Javascript

JQuery

AJAX

XSD

Python

Matlab

R Scripts

Weka





Cost Sensitive Classifier Random Forest Java in weka. The code also does a re-sampling to the training data in the cross validation.
java code
package wekatest;

import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import weka.classifiers.CostMatrix;
import weka.classifiers.Evaluation;
import weka.classifiers.meta.CostSensitiveClassifier;

import weka.classifiers.trees.RandomForest;
import weka.core.Instances;
import weka.core.converters.ArffSaver;
import weka.filters.Filter;
import weka.filters.supervised.instance.Resample;
import weka.filters.unsupervised.instance.Randomize;

/**
*
* @author samy
*/
public class WekaTestRFResampleCost {

/**
* @param args the command line arguments
*/
public static void main(String[] args) throws Exception {
BufferedReader br = null;
double numFolds = 10.0d;
double precisionOne = 0.0d;
double recallOne = 0.0d;
double fmeansureOne = 0.0d;
double precisionTwo = 0.0d;
double recallTwo = 0.0d;
double fmeansureTwo = 0.0d;



double ROCone = 0.0d;
double ROCtwo = 0.0d;



double PRCone = 0.0d;
double PRCtwo = 0.0d;




br = new BufferedReader(new FileReader("D:\\CostResample\\eclipse.arff"));
ArffSaver saverTets = new ArffSaver();
ArffSaver saverTraining = new ArffSaver();
Instances trainData = new Instances(br);
trainData.setClassIndex(trainData.numAttributes() - 1);
br.close();
Randomize randFilterMain = new Randomize();
randFilterMain.setInputFormat(trainData);
trainData = Filter.useFilter(trainData, randFilterMain);
int size = (int) (trainData.numInstances() / numFolds);
int begin = 0; // is index if flod.
int end = size - 1; // is index

System.out.println("Total Size of instances" + trainData.numInstances() + " , flod size=" + size);
for (int i = 1; i <= numFolds; i++) {
System.out.println("Iteration # " + i + " Begin =" + begin + " , end=" + end);
Instances tempTraining = new Instances(trainData);
Instances tempTesting = new Instances(trainData, begin, (end - begin));
for (int j = 0; j < (end - begin); j++) {
tempTraining.delete(begin);
}

//// Filters
Resample resample = new Resample();

resample.setBiasToUniformClass(0.5f);
resample.setInvertSelection(false);
resample.setNoReplacement(false);

resample.setRandomSeed(1);
//smoteFilter.setClassValue("2");
resample.setInputFormat(tempTraining);

System.out.println("Number of instances before filter " + tempTraining.numInstances());

Instances resmapleTempTraining = Filter.useFilter(tempTraining, resample);



System.out.println("Number of instances after filter " + resmapleTempTraining.numInstances());

RandomForest randomForest = new RandomForest();
randomForest.setNumTrees(100);

System.out.println("Started building the model #" + i);
// randomForest.buildClassifier(resmapleTempTraining);

CostSensitiveClassifier costSensitiveClassifier = new CostSensitiveClassifier();
CostMatrix costMatrix = new CostMatrix(2);
// costMatrix.setCell(0, 0, 0.8d);
// costMatrix.setCell(0, 1, 5.0d);
costMatrix.setCell(1, 0, 2d);
// costMatrix.setCell(1, 1, 1.0d);

costSensitiveClassifier.setClassifier(randomForest);
costSensitiveClassifier.setCostMatrix(costMatrix);
costSensitiveClassifier.buildClassifier(resmapleTempTraining);

saverTraining.setInstances(resmapleTempTraining);
saverTraining.setFile(new File("D:\\SumCost\\eclipse\\" + i + "_training.arff"));
saverTets.setInstances(tempTesting);
saverTets.setFile(new File("D:\\SumCost\\eclipse\\" + i + "_testing.arff"));

saverTraining.writeBatch();
saverTets.writeBatch();


System.out.println("Done with building the model");

Evaluation evaluation = new Evaluation(tempTesting);

evaluation.evaluateModel(costSensitiveClassifier, tempTesting);

System.out.println("Results For Class -1- ");
System.out.println("Precision= " + evaluation.precision(0));
System.out.println("Recall= " + evaluation.recall(0));
System.out.println("F-measure= " + evaluation.fMeasure(0));
System.out.println("ROC= " + evaluation.areaUnderROC(0));
System.out.println("Results For Class -2- ");
System.out.println("Precision= " + evaluation.precision(1));
System.out.println("Recall= " + evaluation.recall(1));
System.out.println("F-measure= " + evaluation.fMeasure(1));
System.out.println("ROC= " + evaluation.areaUnderROC(1));
precisionOne += evaluation.precision(0);
recallOne += evaluation.recall(0);
fmeansureOne += evaluation.fMeasure(0);
precisionTwo += evaluation.precision(1);
recallTwo += evaluation.recall(1);
fmeansureTwo += evaluation.fMeasure(1);



ROCone += evaluation.areaUnderROC(0);
ROCtwo += evaluation.areaUnderROC(1);

PRCone += evaluation.areaUnderPRC(0);
PRCtwo += evaluation.areaUnderPRC(1);


begin = end + 1;
end += size;
if (i == (numFolds - 1)) {
end = trainData.numInstances();
}
}




System.out.println("####################################################");
System.out.println("Results For Class -1- ");
System.out.println("Precision= " + precisionOne / numFolds);
System.out.println("Recall= " + recallOne / numFolds);
System.out.println("F-measure= " + fmeansureOne / numFolds);
System.out.println("ROC= " + ROCone / numFolds);
System.out.println("PRC= " + PRCone / numFolds);



System.out.println("Results For Class -2- ");
System.out.println("Precision= " + precisionTwo / numFolds);
System.out.println("Recall= " + recallTwo / numFolds);
System.out.println("F-measure= " + fmeansureTwo / numFolds);
System.out.println("ROC= " + ROCtwo / numFolds);
System.out.println("PRC= " + PRCtwo / numFolds);



}
}




_________________
M. S. Rakha, Ph.D.
Queen's University
Canada


Author:
Mastermind
User avatar Posts: 2715
Have thanks: 74 time

updated.


_________________
Sami
PHD student - SAIL - School Of Computing
Queens' University
Canada


Author:
Site Admin
User avatar Posts: 33
Have thanks: 1 time
Post new topic Reply to topic  [ 2 posts ] 

  Related Posts  to : Cost Sensitive Classifier Random Forest Java in weka
 Weka java code for Random Forest Cross Validation     -  
 random forest algorithm classifier     -  
 Get the important variables of random forest classifier     -  
 Random Search for tuning classifier parameters     -  
 Random Forest Classification (Binary )- Supervised Learning     -  
 KFold Cross-validation Random Forest Binary Classification     -  
 Generating Random Number in java     -  
 calculates the total cost with javascript     -  
 naive Bayes classifier in MATLAB     -  
 php Random image     -  



Topic Tags

Weka Classifiers






Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
All copyrights reserved to codemiles.com 2007-2011
mileX v1.0 designed by codemiles team
Codemiles.com is a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to Amazon.com