/// <summary>
/// Initialize an instance of the evaluator with a training set and a Weka classifier.
/// </summary>
/// <param name="trainingSetPath">The path of the ARFF file to be used as the training set.</param>
/// <param name="classifier">The Weka classifer to be used during evaluation.</param>
public WekaClassification(string trainingSetPath, Classifier classifier)
{
//Load Training Set from file
OriginalDataset = new ConverterUtils.DataSource(trainingSetPath).getDataSet();
OriginalDataset.setClassIndex(OriginalDataset.numAttributes() - 1);
//Assign classifier
Classifier = classifier;
}
public static double EvaluateClassifier(WekaClassifier wekaClassifier, string datasetPath)
{
//classifier.i
Instances dataset = new ConverterUtils.DataSource(datasetPath).getDataSet();
dataset.setClassIndex(dataset.numAttributes() - 1);
if (wekaClassifier.AttributesToRemove != null && wekaClassifier.AttributesToRemove.Length != 0)
{
dataset = GetReducedDataset(dataset, wekaClassifier.AttributesToRemove, null);
}
double quality = 0;
//evaluate classifier performance
Evaluation eval = new Evaluation(dataset);
eval.evaluateModel(wekaClassifier.Classifier, dataset);
//return result based on evaluation
quality = eval.correct() / dataset.numInstances(); //TODO: I'm assuming that this is accuracy. If not please tell me, and I can modify it.
return(quality);
}
public static void ClassifyTest()
{
try
{
weka.core.Instances insts = new ConverterUtils.DataSource("weka").getDataSet();
insts.setClassIndex(insts.numAttributes() - 1);
weka.classifiers.Classifier cl = new weka.classifiers.trees.J48();
Console.WriteLine("Performing " + percentSplit + "% split evaluation.");
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom = new weka.filters.unsupervised.instance.Randomize();
myRandom.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, myRandom);
int trainSize = insts.numInstances()*percentSplit/100;
int testSize = insts.numInstances() - trainSize;
weka.core.Instances train = new weka.core.Instances(insts, 0, trainSize);
cl.buildClassifier(train);
int numCorrect = 0;
for (int i = trainSize; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = cl.classifyInstance(currentInst);
if (predictedClass == insts.instance(i).classValue())
numCorrect++;
}
Console.WriteLine(numCorrect + " out of " + testSize + " correct (" +
(double) ((double) numCorrect/(double) testSize*100.0) + "%)");
}
catch (java.lang.Exception ex)
{
ex.printStackTrace();
}
}
/// <summary>
/// Evaluates the quality of the classifier on a data set post reduction.
/// </summary>
/// <param name="attributesToRemove">The 0-based indices of the attributes to remove from the data set.</param>
/// <param name="instancesToRemove">The 0-based indicies of the instances to remove from the data set.</param>
/// <returns>Accuracy of the classifier post evaluation.</returns>
public static Instances GetReducedDataset(Instances dataset, int[] attributesToRemove, int[] instancesToRemove)
{
#region Validation
if (dataset == null)
{
throw new Exception("Class has not been properly initialized. Please reconstruct the class and try again.");
}
#endregion
//build attributes filter
string attributesFilterString = "";
if (attributesToRemove != null && attributesToRemove.Length > 0)
{
attributesFilterString = "-R ";
for (int i = 0; i < attributesToRemove.Length; i++)
{
attributesFilterString += (attributesToRemove[i] + 1).ToString();
if (i < attributesToRemove.Length - 1)
{
attributesFilterString += ",";
}
}
}
//build instances filter
string instancesFilterString = "";
if (instancesToRemove != null && instancesToRemove.Length > 0)
{
instancesFilterString = "-R ";
for (int i = 0; i < instancesToRemove.Length; i++)
{
instancesFilterString += (instancesToRemove[i] + 1).ToString();
if (i < instancesToRemove.Length - 1)
{
instancesFilterString += ",";
}
}
}
//get reduced set
Instances reducedDataset = new ConverterUtils.DataSource(dataset).getDataSet();//OriginalDataset;
reducedDataset.setClassIndex(reducedDataset.numAttributes() - 1);
if (!String.IsNullOrWhiteSpace(attributesFilterString))
{
Remove attributesFilter = new Remove();
attributesFilter.setOptions(Utils.splitOptions(attributesFilterString));
attributesFilter.setInputFormat(reducedDataset);
reducedDataset = Filter.useFilter(reducedDataset, attributesFilter);
}
if (!String.IsNullOrWhiteSpace(instancesFilterString))
{
weka.filters.unsupervised.instance.RemoveRange instancesFilter = new weka.filters.unsupervised.instance.RemoveRange();
instancesFilter.setOptions(Utils.splitOptions(instancesFilterString));
instancesFilter.setInputFormat(reducedDataset);
reducedDataset = Filter.useFilter(reducedDataset, instancesFilter);
}
return(reducedDataset);
}
