public void Test()
{
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader("D:\\android_analysis\\attributes.arff"));
insts.setClassIndex(insts.numAttributes() - 1);
weka.classifiers.Classifier cl = new weka.classifiers.trees.J48();
cl.buildClassifier(insts);
weka.filters.Filter myRandom = new weka.filters.unsupervised.instance.Randomize();
myRandom.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, myRandom);
int trainSize = (int)(insts.numInstances() * 0.66);
int testSize = insts.numInstances() - trainSize;
weka.core.Instances train = new weka.core.Instances(insts, 0, trainSize);
cl.buildClassifier(train);
for (int i = trainSize; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = cl.classifyInstance(currentInst);
double[] distrs = cl.distributionForInstance(currentInst);
string actual = insts.classAttribute().value((int)currentInst.classValue());
string predicted = insts.classAttribute().value((int)predictedClass);
System.Console.WriteLine("ID: " + (i + 1) + ", " + actual + " --> " + predicted);
}
}
static public int step_test()
{
predict_numCorrect = 0;
double predictedClass;
double actualClass;
listResult = new List <Weka_EachResult>();
weka.core.Instance currentInst;
for (int i = split_trainSize; i < insts.numInstances(); i++)
{
currentInst = insts.instance(i);
predictedClass = classifier.classifyInstance(currentInst);
actualClass = insts.instance(i).classValue();
if (predictedClass == actualClass)
{
predict_numCorrect++;
}
//==================================
listResult.Add(new Weka_EachResult()
{
ID = i.ToString(),
Actual = actualClass.ToString(),
Predict = predictedClass.ToString(),
Diff = (predictedClass - actualClass).ToString()
});
}
//currentInst = insts.lastInstance();
//TheSys.showError(currentInst.ToString(), true);
return(predict_numCorrect);
}
public static void Main(String[] args)
{
try
{
// Load the model
java.io.ObjectInputStream stream = new java.io.ObjectInputStream(new java.io.FileInputStream("iris_j48.model"));
weka.classifiers.Classifier qhClassifier = (weka.classifiers.Classifier)stream.readObject();
stream.close();
// This model was trained on 66% of instances from the iris dataset. Test the model on remaining 34% instances.
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader("iris.arff"));
insts.setClassIndex(insts.numAttributes() - 1);
int percentSplit = 66;
int trainSize = insts.numInstances() * percentSplit / 100;
int testSize = insts.numInstances() - trainSize;
int numCorrect = 0;
for (int i = trainSize; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = qhClassifier.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 e)
{
e.printStackTrace();
}
}
public static void FilterInstances(weka.core.Instances allInstances)
{
DateTime nextHpDate = DateTime.MinValue;
java.util.LinkedList deleteList = new java.util.LinkedList();
for (int i = 0; i < allInstances.numInstances(); ++i)
{
DateTime nowDate = WekaUtils.GetDateValueFromInstances(allInstances, 0, i);
if (TestParameters2.RealTimeMode && i == allInstances.numInstances() - 1)
{
allInstances.instance(i).setClassValue(0);
allInstances.instance(i).setValue(1, WekaUtils.GetTimeFromDate(Parameters.MaxDate) * 1000);
}
else
{
if (nowDate < nextHpDate)
{
deleteList.Add(allInstances.instance(i));
}
else
{
DateTime hpDate = WekaUtils.GetDateValueFromInstances(allInstances, 1, i);
nextHpDate = hpDate;
}
}
}
allInstances.removeAll(deleteList);
}
public void test_has_class_value_with_limited_training_set()
{
TestingRow5[] rows = new[] {
new TestingRow5 {
CLASS = 1.0, ATT_1 = "1"
},
new TestingRow5 {
CLASS = 33.0, ATT_1 = "2"
},
new TestingRow5 {
CLASS = 22.0, ATT_1 = "3"
},
new TestingRow5 {
CLASS = 33.0, ATT_1 = "4"
},
new TestingRow5 {
CLASS = 33.0, ATT_1 = "5"
},
new TestingRow5 {
CLASS = 11.0, ATT_1 = "2"
}
};
InstancesBuilder <TestingRow5> builder = new InstancesBuilder <TestingRow5>(rows, 0, 3);
weka.core.Instances instances = builder.Build();
Assert.AreEqual(3, instances.numAttributes());
Assert.AreEqual(6, instances.numInstances());
CollectionAssert.AreEqual(new[] { "1", "2", "3", "4", "5", "2" }, instances.GetAttrStrings(1));
CollectionAssert.AreEqual(new[] { "0", "1", "0", "0", "0", "1" }, instances.GetAttrStrings(2));
}
public void WriteFile(List <List <string> > numericDataset, string file, List <string> atrNames, List <string> targetValues, bool isTargetNumeric)
{
weka.core.FastVector targetVals = new weka.core.FastVector();
weka.core.Instances dataRel;
for (int i = 0; i < targetValues.Count; i++)
{
targetVals.addElement(targetValues[i]);
}
weka.core.Instances data;
weka.core.FastVector atts = new weka.core.FastVector();
// fill and prepare the dataset for the arrf file
for (int j = 0; j < insts.numAttributes(); j++)
{
if (j == insts.numAttributes() - 1 && isTargetNumeric == false) // target value can be nominal
{
atts.addElement(new weka.core.Attribute(atrNames[j], targetVals));
}
else
{
atts.addElement(new weka.core.Attribute(atrNames[j]));
}
}
data = new weka.core.Instances("MyRelation", atts, 0);
for (int i = 0; i < insts.numInstances(); i++)
{
double[] vals = new double[insts.numAttributes()];
for (int j = 0; j < insts.numAttributes(); j++)
{
if (j == insts.numAttributes() - 1 && isTargetNumeric == false) // target value can be nominal
{
vals[j] = targetVals.indexOf(numericDataset[j][i]);
}
else
{
vals[j] = Convert.ToDouble(numericDataset[j][i]);
}
}
data.add(new weka.core.DenseInstance(1.0, vals));
}
if (File.Exists(file))
{
File.Delete(file);
}
var saver = new weka.core.converters.ArffSaver();
saver.setInstances(data);
saver.setFile(new java.io.File(file));
// files are saved into {AppFolder}/bin/Debug folder. You can find two files in this path.
saver.writeBatch();
}
private void btnLoadStep_Click(object sender, EventArgs e)
{
if (m_loadStepInstances == null)
{
using (OpenFileDialog d = new OpenFileDialog())
{
d.Filter = "Arff File|*.arff";
if (d.ShowDialog() == System.Windows.Forms.DialogResult.OK)
{
m_loadStepInstances = new weka.core.Instances(new java.io.BufferedReader(new java.io.FileReader(d.FileName)));
m_loadStepInstances.setClassIndex(m_loadStepInstances.numAttributes() - 1);
clear_all();
}
}
}
else
{
for (int i = m_loadStepIdx; i < m_loadStepInstances.numInstances(); ++i)
{
var ins = m_loadStepInstances.instance(i);
var p = new valuePoint(ins.value(0), ins.value(1), (int)ins.classValue());
if (p.x < 0 || p.x >= 1 || p.y < 0 || p.y >= 1)
{
continue;
}
point_list.Add(p);
draw_point(p);
m_loadStepIdx = i + 1;
if (i % 1000 == 0)
{
break;
}
}
pictureBox1.Invalidate();
if (m_loadStepIdx == m_loadStepInstances.numInstances())
{
m_loadStepIdx = 0;
m_loadStepInstances = null;
}
}
}
//Knn
public static double Knn(weka.core.Instances insts)
{
try
{
insts.setClassIndex(insts.numAttributes() - 1);
Knncl = new weka.classifiers.lazy.IBk();
weka.filters.Filter myDummy = new weka.filters.unsupervised.attribute.NominalToBinary();
myDummy.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, myDummy);
weka.filters.Filter myNormalize = new weka.filters.unsupervised.instance.Normalize();
myNormalize.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, myNormalize);
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);
Knncl.buildClassifier(train);
int numCorrect = 0;
for (int i = trainSize; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = Knncl.classifyInstance(currentInst);
if (predictedClass == insts.instance(i).classValue())
{
numCorrect++;
}
}
return((double)numCorrect / (double)testSize * 100.0);
}
catch (java.lang.Exception ex)
{
ex.printStackTrace();
return(0);
}
}
//private weka.core.Instances m_instances;
public void evaluateModel(double[] v, weka.core.Instances instances)
{
double[] c = new double[instances.numInstances()];
for (int i = 0; i < c.Length; ++i)
{
c[i] = instances.instance(i).classValue();
}
evaluateModel(v, c);
}
//Only 1 output: last instance
static public string do_Classification_bySerialClassfier_1out_standAlone
(SerializedClassifier serialClassifier, weka.core.Instances instances, int colClass)
{
instances.setClassIndex(colClass);
weka.core.Instance each = instances.instance(instances.numInstances() - 1);
double predictedClass = serialClassifier.classifyInstance(each);
return(instances.classAttribute().value((int)predictedClass));
}
public static string classifyTest()
{
try
{
String result = "";
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader("C:\\Program Files\\Weka-3-7\\data\\iris.arff"));
insts.setClassIndex(insts.numAttributes() - 1);
weka.classifiers.Classifier cl = new weka.classifiers.trees.J48();
// Console.WriteLine("Performing " + percentSplit + "% split evaluation.");
result += "Performing " + percentSplit + "% split evaluation.\n";
//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) + "%)");
result += (numCorrect + " out of " + testSize + " correct (" + (double)((double)numCorrect / (double)testSize * 100.0) + "%)");
return(result);
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
return(ex.Message);
}
}
public static void classifyTest()
{
try
{
Console.WriteLine("Hello Java, from C#!");
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader("D:/iris.arff"));
insts.setClassIndex(insts.numAttributes() - 1);
weka.classifiers.Classifier cl = new weka.classifiers.trees.J48();
Console.WriteLine("Performing " + 33 + "% 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() * 33 / 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 (Exception ex)
{
}
}
public static void CalculateSuccessForAnn(weka.core.Instances originalInsts)
{
try
{
var form = Form.ActiveForm as Form1;
form.successPrcAnn.Text = "Training...";
form.successRtAnn.Text = "../" + testSize;
weka.core.Instances insts = originalInsts;
// Pre-process
insts = ConvertNominalToNumeric(insts);
insts = Normalize(insts);
// Classify
weka.classifiers.Classifier cl = new weka.classifiers.functions.MultilayerPerceptron();
weka.core.Instances train = new weka.core.Instances(insts, 0, trainSize);
cl.buildClassifier(train);
int numCorrect = 0;
double percentage = 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++;
}
percentage = (double)numCorrect / (double)testSize * 100.0;
form.successRtAnn.Text = numCorrect + "/" + testSize;
form.successPrcAnn.Text = String.Format("{0:0.00}", percentage) + "%";
}
succesRates.Add(Classifier.ANN, percentage);
classifiers.Add(Classifier.ANN, cl);
}
catch (java.lang.Exception ex)
{
ex.printStackTrace();
MessageBox.Show(ex.ToString(), "Error for Neural Network", MessageBoxButtons.OK, MessageBoxIcon.Error);
}
catch (Exception)
{
MessageBox.Show("Error for Neural Network", "Error for Neural Network", MessageBoxButtons.OK, MessageBoxIcon.Error);
}
}
public String PrintClassifierTestReport()
{
try {
Debug.Log("Classifier: Number of instances: " + playerData.numInstances());
weka.classifiers.Evaluation eval = new weka.classifiers.Evaluation(playerData);
eval.crossValidateModel(classifier, playerData, 10, new java.util.Random(1));
Debug.Log(eval.toSummaryString("\nClassifier: Cross Validate Results: \n======\n", false));
return(eval.toSummaryString("\nResults\n======\n", false));
}
catch (java.lang.Exception ex)
{
Debug.LogError(ex.getMessage());
}
return(null);
}
public void Test2()
{
java.io.ObjectInputStream ois = new java.io.ObjectInputStream(new java.io.FileInputStream("D:\\android_analysis\\som_model.model"));
weka.classifiers.Classifier cl = (weka.classifiers.Classifier)ois.readObject();
ois.close();
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader("D:\\android_analysis\\test1.arff"));
insts.setClassIndex(insts.numAttributes() - 1);
for (int i = 0; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = cl.classifyInstance(currentInst);
double[] distrs = cl.distributionForInstance(currentInst);
//string actual = insts.classAttribute().value((int)currentInst.classValue());
//string predicted = insts.classAttribute().value((int)predictedClass);
// System.Console.WriteLine("ID: " + (i + 1) + ", " + predicted);
}
}
public void test_ignore_attributes()
{
TestingRow6[] rows = new[] {
new TestingRow6 {
CLASS = 1.0, ATT_1 = "1", ATT_2 = "1.1", ATT_3 = "1.2"
},
new TestingRow6 {
CLASS = 2.0, ATT_1 = "2", ATT_2 = "2.1", ATT_3 = "2.2"
}
};
InstancesBuilder <TestingRow6> builder = new InstancesBuilder <TestingRow6>(rows, 0);
weka.core.Instances instances = builder.Build();
Assert.AreEqual(3, instances.numAttributes()); // 1 is ignored
Assert.AreEqual(2, instances.numInstances());
CollectionAssert.AreEqual(new[] { "1", "2" }, instances.GetAttrStrings(1));
CollectionAssert.AreEqual(new[] { "1.2", "2.2" }, instances.GetAttrStrings(2));
}
public void flatten_attribute()
{
FlattenClass[] rows = new[] {
new FlattenClass {
CLASS = 0, Att1 = new [] { 1, 2, 3, 4, 5 }
},
new FlattenClass {
CLASS = 1, Att1 = new [] { 6, 7, 8, 9, 10 }
}
};
InstancesBuilder <FlattenClass> builder = new InstancesBuilder <FlattenClass>(rows, 0);
weka.core.Instances instances = builder.Build();
Assert.AreEqual(2, instances.numInstances());
Assert.AreEqual(6, instances.numAttributes());
CollectionAssert.AreEqual(new[] { "1", "2" }, instances.GetAttrStrings(1));
CollectionAssert.AreEqual(new[] { "1.2", "2.2" }, instances.GetAttrStrings(2));
}
static public List <string> do_Classification_bySerialClassfier_standAlone
(SerializedClassifier serialClassifier, weka.core.Instances instances, int colClass)
{
instances.setClassIndex(colClass);
List <string> listPredictClass = new List <string>();
double predictedClass; double actualClass; predict_numCorrect = 0;
weka.core.Instance each;
for (int i = 0; i < instances.numInstances(); i++)
{
each = instances.instance(i);
predictedClass = serialClassifier.classifyInstance(each);
actualClass = instances.instance(i).classValue();
if (predictedClass == actualClass)
{
predict_numCorrect++;
}
//Get Class Name
listPredictClass.Add(instances.classAttribute().value((int)predictedClass));
}
return(listPredictClass);
}
public override double classifyInstance(weka.core.Instance instance)
{
if (m_instances.numInstances() == 0)
{
return(2);
}
if (m_instances.numAttributes() != instance.numAttributes())
{
throw new AssertException("different attribute.");
}
int n = instance.numAttributes();
List <Tuple <int, int> > dist = new List <Tuple <int, int> >();
for (int i = 0; i < m_instances.numInstances(); ++i)
{
int d1 = 0, d2 = 0;
weka.core.Instance instanceI = m_instances.instance(i);
for (int j = 0; j < n; ++j)
{
//d += (int)((instanceI.value(j) - instance.value(j)) * (instanceI.value(j) - instance.value(j)));
if (instanceI.value(j) != instance.value(j))
{
d1++;
}
if (instance.value(j) != 0)
{
d2++;
}
}
int c = (int)instanceI.classValue();
dist.Add(new Tuple <int, int>(d1, c));
}
if (dist.Count == 0)
{
return(2);
}
dist.Sort(new Comparison <Tuple <int, int> >((x, y) =>
{
return(x.Item1.CompareTo(y.Item1));
}));
int sum = 0, count = 0;
for (int i = 0; i < dist.Count; ++i)
{
if (dist[i].Item1 < 4)
{
sum += dist[i].Item2;
count++;
}
else
{
break;
}
}
if (count == 0)
{
return(2);
}
if (count < m_instances.numInstances() / 70)
{
return(2);
}
return((int)Math.Round((double)sum / count));
}
private void button1_Click(object sender, EventArgs e)
{
string fname = "";
OpenFileDialog dialog = new OpenFileDialog();
dialog.Filter =
"Weka Files (*.arff)|*.arff|All files (*.*)|*.*";
dialog.InitialDirectory = Application.StartupPath;
dialog.Title = "Select a .arff file";
if (dialog.ShowDialog() == DialogResult.OK)
{
fname = dialog.FileName;
//label5.Text = System.IO.Directory.;
}
if (fname == "")
{
return;
}
try
{
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader(fname.ToString()));
insts.setClassIndex(insts.numAttributes() - 1);
Classifier cl = new weka.classifiers.functions.SMO();
//label1.Text = "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++;
}
}
//label1.Text = numCorrect + " out of " + testSize + " correct (" +
//(double)((double)numCorrect / (double)testSize * 100.0) + "%)";
label6.Text = testSize.ToString();
label7.Text = numCorrect.ToString();
label8.Text = (double)((double)numCorrect / (double)testSize * 100.0) + "%";
double result_perc = (double)((double)numCorrect / (double)testSize * 100.0);
result_perc = Math.Truncate(result_perc);
try
{
// Send Data On Serial port
SerialPort serialPort = new SerialPort("COM" + textBox1.Text + "", Int32.Parse(textBox2.Text), Parity.None, 8);
serialPort.Open();
if (result_perc <= 75)
{
serialPort.WriteLine("1");
}
serialPort.WriteLine("a");
serialPort.Close();
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
}
}
catch (java.lang.Exception ex)
{
MessageBox.Show(ex.getMessage().ToString(), "");
}
}
public double GetProb(weka.core.Instances instances)
{
x = 0; y = 0; z = 0;
for (int i = 0; i < instances.numInstances(); ++i)
{
double v = instances.instance(i).classValue();
if (v == 0)
{
x++;
}
else if (v == 1)
{
y++;
}
else if (v == 2)
{
z++;
}
else if (v == 3)
{
w++;
}
else
{
throw new ArgumentException("invalid v");
}
}
int n = x + y + z + w;
double tp = (TestParameters2.tpStart + 1 + TestParameters2.tpCount) / 2.0;
double sl = (TestParameters2.slStart + 1 + TestParameters2.slCount) / 2.0;
double r1 = (x * tp - y * sl + w / 2 * tp - w / 2 * sl) * x
+ (y * tp - x * sl + w / 2 * tp - w / 2 * sl) * y
- (x * sl + y * sl + w * sl) * z
+ tp * w * n;
r1 = r1 / n / n;
double r;
if (x > y)
{
r = x * tp - y * sl - z * sl + w * tp;
}
else if (x < y)
{
r = -x * sl + y * tp - z * sl + w * tp;
}
else
{
r = 0;
}
r = r / n;
//DateTime date = WekaUtils.GetDateTimeValueFromInstances(instances, 0, 0);
//DateTime hpdate = WekaUtils.GetDateTimeValueFromInstances(instances, 1, 0);
//using (StreamWriter sw = new StreamWriter("d:\\p.txt", true))
//{
// sw.Write(string.Format("{2}, {3}, {4}, {5}, ",
// date.ToString(Parameters.DateTimeFormat), hpdate.ToString(Parameters.DateTimeFormat),
// x, y, z, w, r));
//}
return(r1);
}
// Test the classification result of each map that a user played,
// with the data available as if they were playing through it
public static void classifyTest(String dataString, String playerID)
{
String results = "";
try {
java.io.StringReader stringReader = new java.io.StringReader(dataString);
java.io.BufferedReader buffReader = new java.io.BufferedReader(stringReader);
/* NOTE THAT FOR NAIVE BAYES ALL WEIGHTS CAN BE = 1*/
//weka.core.converters.ConverterUtils.DataSource source = new weka.core.converters.ConverterUtils.DataSource("iris.arff");
weka.core.Instances data = new weka.core.Instances(buffReader); //source.getDataSet();
// setting class attribute if the data format does not provide this information
// For example, the XRFF format saves the class attribute information as well
if (data.classIndex() == -1)
{
data.setClassIndex(data.numAttributes() - 1);
}
weka.classifiers.Classifier cl;
for (int i = 3; i < data.numInstances(); i++)
{
cl = new weka.classifiers.bayes.NaiveBayes();
//cl = new weka.classifiers.trees.J48();
//cl = new weka.classifiers.lazy.IB1();
//cl = new weka.classifiers.functions.MultilayerPerceptron();
((weka.classifiers.functions.MultilayerPerceptron)cl).setHiddenLayers("12");
weka.core.Instances subset = new weka.core.Instances(data, 0, i);
cl.buildClassifier(subset);
weka.classifiers.Evaluation eval = new weka.classifiers.Evaluation(subset);
eval.crossValidateModel(cl, subset, 3, new java.util.Random(1));
results = results + eval.pctCorrect(); // For accuracy measurement
/* For Mathews Correlation Coefficient */
//double TP = eval.numTruePositives(1);
//double FP = eval.numFalsePositives(1);
//double TN = eval.numTrueNegatives(1);
//double FN = eval.numFalseNegatives(1);
//double correlationCoeff = ((TP*TN)-(FP*FN))/Math.Sqrt((TP+FP)*(TP+FN)*(TN+FP)*(TN+FN));
//results = results + correlationCoeff;
if (i != data.numInstances() - 1)
{
results = results + ", ";
}
if (i == data.numInstances() - 1)
{
Debug.Log("Player: " + playerID + ", Num Maps: " + data.numInstances() + ", AUC: " + eval.areaUnderROC(1));
}
}
} catch (java.lang.Exception ex)
{
Debug.LogError(ex.getMessage());
}
// Write values to file for a matlab read
// For accuracy
StreamWriter writer = new StreamWriter("DataForMatlab/" + playerID + "_CrossFoldValidations_NeuralNet.txt");
//StreamWriter writer = new StreamWriter("DataForMatlab/"+playerID+"_CrossFoldCorrCoeff.txt"); // For mathews cc
writer.WriteLine(results);
writer.Close();
Debug.Log(playerID + " has been written to file");
}
/// <summary> Calculates the area under the ROC curve. This is normalised so
/// that 0.5 is random, 1.0 is perfect and 0.0 is bizarre.
///
/// </summary>
/// <param name="tcurve">a previously extracted threshold curve Instances.
/// </param>
/// <returns> the ROC area, or Double.NaN if you don't pass in
/// a ThresholdCurve generated Instances.
/// </returns>
public static double getROCArea(Instances tcurve)
{
//UPGRADE_NOTE: Final was removed from the declaration of 'n '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
int n = tcurve.numInstances();
if (!RELATION_NAME.Equals(tcurve.relationName()) || (n == 0))
{
return System.Double.NaN;
}
//UPGRADE_NOTE: Final was removed from the declaration of 'tpInd '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
int tpInd = tcurve.attribute(TRUE_POS_NAME).index();
//UPGRADE_NOTE: Final was removed from the declaration of 'fpInd '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
int fpInd = tcurve.attribute(FALSE_POS_NAME).index();
//UPGRADE_NOTE: Final was removed from the declaration of 'tpVals '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double[] tpVals = tcurve.attributeToDoubleArray(tpInd);
//UPGRADE_NOTE: Final was removed from the declaration of 'fpVals '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double[] fpVals = tcurve.attributeToDoubleArray(fpInd);
//UPGRADE_NOTE: Final was removed from the declaration of 'tp0 '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double tp0 = tpVals[0];
//UPGRADE_NOTE: Final was removed from the declaration of 'fp0 '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double fp0 = fpVals[0];
double area = 0.0;
//starts at high values and goes down
double xlast = 1.0;
double ylast = 1.0;
for (int i = 1; i < n; i++)
{
//UPGRADE_NOTE: Final was removed from the declaration of 'x '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double x = fpVals[i] / fp0;
//UPGRADE_NOTE: Final was removed from the declaration of 'y '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double y = tpVals[i] / tp0;
//UPGRADE_NOTE: Final was removed from the declaration of 'areaDelta '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double areaDelta = (y + ylast) * (xlast - x) / 2.0;
/*
System.err.println("[" + i + "]"
+ " x=" + x
+ " y'=" + y
+ " xl=" + xlast
+ " yl=" + ylast
+ " a'=" + areaDelta);
*/
area += areaDelta;
xlast = x;
ylast = y;
}
//make sure ends at 0,0
if (xlast > 0.0)
{
//UPGRADE_NOTE: Final was removed from the declaration of 'areaDelta '. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1003'"
double areaDelta = ylast * xlast / 2.0;
//System.err.println(" a'=" + areaDelta);
area += areaDelta;
}
//System.err.println(" area'=" + area);
return area;
}
public bool PrepareDataset()
{
try
{
weka.filters.Filter missingFilter = new weka.filters.unsupervised.attribute.ReplaceMissingValues(); // missing values handled
missingFilter.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, missingFilter);
bool isTargetNumeric = insts.attribute(insts.numAttributes() - 1).isNumeric();
List <bool> isNumeric = new List <bool>();
List <bool> is2Categorical = new List <bool>();
List <List <string> > numericColumns = new List <List <string> >();
List <string> atrNames = new List <string>();
for (int i = 0; i < insts.numAttributes(); i++)
{
atrNames.Add(insts.attribute(i).name());
bool isNum = insts.attribute(i).isNumeric();
isNumeric.Add(isNum);
if (isNum == true)
{
numericColumns.Add(new List <string>());
for (int j = 0; j < insts.numInstances(); j++)
{
numericColumns[numericColumns.Count - 1].Add(insts.instance(j).toString(i));
}
}
}
weka.filters.unsupervised.attribute.Discretize myDiscretize = new weka.filters.unsupervised.attribute.Discretize();
myDiscretize.setInputFormat(insts);
myDiscretize.setFindNumBins(true);
insts = weka.filters.Filter.useFilter(insts, myDiscretize);
List <List <string> > atrs = new List <List <string> >();
for (int i = 0; i < insts.numAttributes(); i++)
{
atrs.Add(new List <string>());
for (int j = 0; j < insts.attribute(i).numValues(); j++)
{
string sub_category = insts.attribute(i).value(j);
string temp = sub_category.Replace("'", string.Empty);
atrs[atrs.Count - 1].Add(temp);
}
if (atrs[atrs.Count - 1].Count == 2)
{
is2Categorical.Add(true);
}
else
{
is2Categorical.Add(false);
}
}
List <List <string> > lst = new List <List <string> >();
for (int i = 0; i < insts.numInstances(); i++)
{
lst.Add(new List <string>());
for (int j = 0; j < insts.instance(i).numValues(); j++)
{
string temp = insts.instance(i).toString(j);
temp = temp.Replace("\\", string.Empty);
temp = temp.Replace("'", string.Empty);
lst[lst.Count - 1].Add(temp);
}
}
List <string> targetValues = atrs[insts.numAttributes() - 1];
List <List <string> > giniDataset = ConvertToNumericWithGini(lst, atrs);
giniDataset = Arrange2CategoricalColumns(giniDataset, lst, is2Categorical);
giniDataset = ChangeBackNumericalColumns(giniDataset, numericColumns, isNumeric);
WriteFile(giniDataset, filename + "-numeric-gini.arff", atrNames, targetValues, isTargetNumeric);
List <List <string> > twoingDataset = ConvertToNumericWithTwoing(lst, atrs);
twoingDataset = Arrange2CategoricalColumns(twoingDataset, lst, is2Categorical);
twoingDataset = ChangeBackNumericalColumns(twoingDataset, numericColumns, isNumeric);
WriteFile(twoingDataset, filename + "-numeric-twoing.arff", atrNames, targetValues, isTargetNumeric);
return(true);
}
catch (Exception e)
{
return(false);
}
}
//public object Clone() {
// MemoryStream ms = new MemoryStream(500000000);
// System.Runtime.Serialization.Formatters.Binary.BinaryFormatter bf =
// new System.Runtime.Serialization.Formatters.Binary.BinaryFormatter(null,
// new System.Runtime.Serialization.StreamingContext(System.Runtime.Serialization.StreamingContextStates.Clone));
// bf.Serialize(ms, this);
// ms.Seek(0, SeekOrigin.Begin);
// object obj = bf.Deserialize(ms);
// ms.Close();
// return obj;
//}
public void train(bool LoadingFromFile)
{
if (!LoadingFromFile) {
DisplayMessage("Begin training on Efigi galaxies...");
Console.Write("Begin training on Efigi galaxies...");
}
else {
DisplayImage(0);
DisplayMessage("Load from file...");
Console.Write("Load from file...");
frV = new GeneralMatrix(ReadFVMatrix(0));
fgV = new GeneralMatrix(ReadFVMatrix(1));
fbV = new GeneralMatrix(ReadFVMatrix(2));
}
weka.classifiers.trees.M5P tree = new weka.classifiers.trees.M5P();
String[] options = new String[1];
weka.core.converters.ConverterUtils.DataSource source = new weka.core.converters.ConverterUtils.DataSource(OutputDir + "Results/" + "resultsGalaxy.arff");
data = source.getDataSet();
if (data == null) {
DisplayMessage("Cannot load from file.");
throw new Exception("Arff File not valid");
}
data.setClassIndex(0);
tree.buildClassifier(data);
StreamWriter output = new StreamWriter(OutputDir + "Results/" + "classification.txt");
rmse = 0.0;
int classifiedCount = 0;
weka.filters.unsupervised.attribute.Remove rm = new weka.filters.unsupervised.attribute.Remove();
rm.setInputFormat(data);
fc = new FilteredClassifier();
fc.setFilter(rm);
fc.setClassifier(tree);
for (int i = 0; i < data.numInstances(); i++) {
int classPrediction = (int)Math.Round(fc.classifyInstance(data.instance(i)));
if (classPrediction < -6) {
classPrediction = -6;
}
else if (classPrediction > 11) {
classPrediction = 11;
}
int actualClass = (int)Math.Round(data.instance(i).classValue());
int error = Math.Abs(classPrediction - actualClass);
rmse += error * error;
classifiedCount++;
output.WriteLine("\n" + classPrediction + ", " + error);
if (i % 10 == 0 && !LoadingFromFile)
DisplayImage(i);
}
rmse = Math.Sqrt(rmse / classifiedCount);
output.WriteLine("\nRMSE: " + rmse);
DisplayMessage("RMSE: " + rmse);
output.Flush();
output.Close();
output.Dispose();
readyToClassify = true;
Console.WriteLine("Finished training on Efigi galaxies; RMSE: " + rmse.ToString());
}
public void classifyTest()
{
try
{
CSV2Arff();
java.io.FileReader arrfFile = new java.io.FileReader("D:/final_version/Gesture-Gis-master/GestureGis2/ComparisonFeaturefile.arff");
weka.core.Instances insts = new weka.core.Instances(arrfFile);
//weka.core.Instances insts2 = new weka.core.Instances(new java.io.FileReader("D:/Gesture-Gis-master/GestureGis2/ComparisonFeaturefile.arff"));
insts.setClassIndex(insts.numAttributes() - 1);
//int percentSplit = 66;
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 count = insts.numInstances();
int trainSize = count - 1;
int testSize = count - trainSize;
weka.core.Instances train = new weka.core.Instances(insts, 0, trainSize);
cl.buildClassifier(train);
//weka.core.Instance current = insts2.instance(0);
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++;
* }*/
int index = count - 1;
weka.core.Instance currentInst = insts.instance(index);
double predictedClass = cl.classifyInstance(currentInst);
int pre = (int)predictedClass;
if (predictedClass == insts.instance(index).classValue())
{
numCorrect++;
}
//insts.instance(index).classAttribute();
//insts.attribute(11);
string s = insts.toString();
s = s.Substring(s.IndexOf("{") + 1);
s = s.Substring(0, s.IndexOf("}"));
s = s.Substring(0, s.Length);
string[] ae = s.Split(',');
/*ArrayList arr = new ArrayList();
* string path_class = @"D:\final_version\Gesture-Gis-master\GestureGis2\Classfile.txt";
* using (StreamReader reader = new StreamReader(path_class))
* {
* while (!reader.EndOfStream)
* {
* arr.Add(reader.ReadLine());
* }
* reader.Close();
* }*/
PredictedClassbyWeka = (string)(ae[pre]);
arrfFile.close();
//insts.instance(index).attribute(3);
/*System.Diagnostics.Debug.WriteLine(numCorrect + " out of " + testSize + " correct (" +
* (double)((double)numCorrect / (double)testSize * 100.0) + "%)");
* Console.WriteLine(numCorrect + " out of " + testSize + " correct (" +
* (double)((double)numCorrect / (double)testSize * 100.0) + "%)");*/
}
catch (java.lang.Exception ex)
{
ex.printStackTrace();
}
}
// Test the classification result of each map that a user played,
// with the data available as if they were playing through it
public static void classifyTest(String dataString, String playerID)
{
try {
java.io.StringReader stringReader = new java.io.StringReader(dataString);
java.io.BufferedReader buffReader = new java.io.BufferedReader(stringReader);
/* NOTE THAT FOR NAIVE BAYES ALL WEIGHTS CAN BE = 1*/
//weka.core.converters.ConverterUtils.DataSource source = new weka.core.converters.ConverterUtils.DataSource("iris.arff");
weka.core.Instances thisData = new weka.core.Instances(buffReader); //source.getDataSet();
if (thisData.classIndex() == -1)
{
thisData.setClassIndex(thisData.numAttributes() - 1);
}
weka.core.Instances thisUniqueData = new weka.core.Instances(thisData);
if (thisUniqueData.classIndex() == -1)
{
thisUniqueData.setClassIndex(thisUniqueData.numAttributes() - 1);
}
thisUniqueData.delete();
if (allUniqueData == null)
{
allUniqueData = new weka.core.Instances(thisData);
if (allUniqueData.classIndex() == -1)
{
allUniqueData.setClassIndex(allUniqueData.numAttributes() - 1);
}
allUniqueData.delete();
}
weka.core.InstanceComparator com = new weka.core.InstanceComparator(false);
for (int i = 0; i < thisData.numInstances(); i++)
{
bool dup = false;
for (int j = 0; j < allUniqueData.numInstances(); j++)
{
if (com.compare(thisData.instance(i), allUniqueData.instance(j)) == 0)
{
Debug.Log("Duplicate found!");
dup = true;
break;
}
}
if (!dup)
{
allUniqueData.add(thisData.instance(i));
}
else
{
dupInstances++;
}
}
for (int i = 0; i < thisData.numInstances(); i++)
{
bool dup = false;
for (int j = 0; j < thisUniqueData.numInstances(); j++)
{
if (com.compare(thisData.instance(i), thisUniqueData.instance(j)) == 0)
{
Debug.Log("Duplicate found!");
dup = true;
break;
}
}
if (!dup)
{
thisUniqueData.add(thisData.instance(i));
}
else
{
dupInstancesSamePlayer++;
}
}
//Debug.Log("All Data Instance Count = " + thisData.numInstances());
//Debug.Log("Unique Data Instance Count = " + thisUniqueData.numInstances());
//Debug.Log("Done!");
} catch (java.lang.Exception ex)
{
Debug.LogError(ex.getMessage());
}
}
public override void buildClassifier(Instances instances)
{
m_mustValue = null;
var weights = MincostLiblinearClassifier.GetCount(instances);
if (weights[0] == 0)
{
m_mustValue = 2;
m_delta = 0;
return;
}
else if (weights[2] == 0)
{
m_mustValue = 0;
m_delta = 0;
return;
}
m_sampleInstances = new Instances(instances, 0);
// can classifier handle the data?
getCapabilities().testWithFail(instances);
Instances trainInstances = new Instances(instances, 0, instances.numInstances());
AddInstancesAccordWeight(trainInstances);
if (System.IO.File.Exists(m_trainingFile))
{
System.IO.File.Delete(m_trainingFile);
}
libsvmSaver.setInstances(trainInstances);
libsvmSaver.setFile(new java.io.File(m_trainingFile));
libsvmSaver.writeBatch();
//ConvertNorminalToString(m_trainingFile);
if (System.IO.File.Exists(m_modelFile))
{
System.IO.File.Delete(m_modelFile);
}
string[] options = Utils.splitOptions(m_trainArgs);
int idx = Utils.getOptionPos('c', options);
if (idx != -1)
{
double c = Convert.ToDouble(options[idx + 1]);
c = c * trainInstances.numInstances() / 100.0;
options[idx + 1] = c.ToString();
m_trainArgs = Utils.joinOptions(options);
}
learner.ExecuteLearner(s_learnerPath, m_trainingFile, m_modelFile, m_trainArgs);
if (!System.IO.File.Exists(m_modelFile))
{
throw new InvalidOperationException(learner.Output);
}
m_modelData = System.IO.File.ReadAllBytes(m_modelFile);
GetBestDelta(instances);
if (System.IO.File.Exists(m_trainingFile))
{
System.IO.File.Delete(m_trainingFile);
}
}
//private void ConvertNorminalToString(string fileName)
//{
// List<string> list = new List<string>();
// using (System.IO.StreamReader sr = new System.IO.StreamReader(fileName))
// {
// while (true)
// {
// if (sr.EndOfStream)
// break;
// string s = sr.ReadLine();
// if (string.IsNullOrEmpty(s))
// continue;
// int idx = s.IndexOf(' ');
// string c = idx == -1 ? s : s.Substring(0, idx);
// if (Convert.ToDouble(c) == 0)
// {
// list.Add("-1.0 " + (idx == -1 ? string.Empty : s.Substring(idx + 1)));
// }
// else if (Convert.ToDouble(c) == 1)
// {
// list.Add("0.0 " + (idx == -1 ? string.Empty : s.Substring(idx + 1)));
// }
// else if (Convert.ToDouble(c) == 2)
// {
// list.Add("+1.0 " + (idx == -1 ? string.Empty : s.Substring(idx + 1)));
// }
// else
// {
// list.Add(s);
// }
// }
// }
// using (System.IO.StreamWriter sw = new System.IO.StreamWriter(fileName))
// {
// foreach (string s in list)
// {
// sw.WriteLine(s);
// }
// }
//}
//private Random m_randomGenerator;
private void AddInstancesAccordWeight(Instances instances)
{
// 0, 2
double[] weights = MincostLiblinearClassifier.GetCount(instances);
if (weights == null)
return;
double c = m_tp / m_sl;
if (c == 1 && weights[0] == weights[1])
return;
int n = 0;
int toCopyClass = 0;
if (c >= 1)
{
int shouldWeight1 = (int)(c * weights[1]);
n = (int)(shouldWeight1 - weights[1]);
toCopyClass = 2;
}
else
{
int shouldShouldWeight0 = (int)(1 / c * weights[0]);
n = (int)(weights[1] - weights[0]);
toCopyClass = 0;
}
//m_randomGenerator = new Random((int)System.DateTime.Now.Ticks);
List<Instance> copyInstances = new List<Instance>();
for (int i = 0; i < instances.numInstances(); ++i)
{
if (instances.instance(i).classValue() == toCopyClass)
{
copyInstances.Add(instances.instance(i));
}
}
int nAll = n / copyInstances.Count;
for (int j = 0; j < nAll; ++j)
{
for (int i = 0; i < copyInstances.Count; ++i)
{
Instance newInstance = new weka.core.DenseInstance(copyInstances[i]);
instances.add(newInstance);
newInstance.setDataset(instances);
}
}
//for (int j = 0; j < n - nAll * copyInstances.Count; ++j)
//{
// int idx = (int)(m_randomGenerator.NextDouble() * copyInstances.Count);
// idx = Math.Min(idx, copyInstances.Count - 1);
// Instance newInstance = new weka.core.DenseInstance(copyInstances[idx]);
// instances.add(newInstance);
// newInstance.setDataset(instances);
//}
if (n - nAll * copyInstances.Count > 0)
{
Instance avgInstance = new weka.core.DenseInstance(instances.numAttributes());
for (int i = 0; i < avgInstance.numAttributes(); ++i)
{
double sum = 0;
for (int j = 0; j < copyInstances.Count; ++j)
{
sum += copyInstances[j].value(i);
}
avgInstance.setValue(i, sum / copyInstances.Count);
}
for (int j = 0; j < n - nAll * copyInstances.Count; ++j)
{
Instance newInstance = new weka.core.DenseInstance(avgInstance);
instances.add(newInstance);
}
}
}
public double[] distributionForInstances(Instances instances)
{
double[] ret = new double[instances.numInstances()];
if (m_mustValue.HasValue)
{
for (int i = 0; i < ret.Length; ++i)
ret[i] = m_mustValue == 0 ? m_delta.Value - 1 : m_delta.Value + 1;
return ret;
}
if (System.IO.File.Exists(m_testFile))
{
System.IO.File.Delete(m_testFile);
}
libsvmSaver.setInstances(instances);
libsvmSaver.setFile(new java.io.File(m_testFile));
libsvmSaver.writeBatch();
//ConvertNorminalToString(m_testFile);
if (System.IO.File.Exists(m_testOutputFile))
{
System.IO.File.Delete(m_testOutputFile);
}
classifier.ExecuteClassifier(s_classifierPath, m_testFile, m_modelFile, m_testOutputFile);
if (!System.IO.File.Exists(m_testOutputFile))
{
throw new InvalidOperationException(classifier.Output);
}
using (System.IO.StreamReader sr = new System.IO.StreamReader(m_testOutputFile))
{
for (int i = 0; i < ret.Length; ++i)
{
string s = sr.ReadLine();
ret[i] = Double.Parse(s);
}
}
if (System.IO.File.Exists(m_testFile))
{
System.IO.File.Delete(m_testFile);
}
if (System.IO.File.Exists(m_testOutputFile))
{
System.IO.File.Delete(m_testOutputFile);
}
return ret;
}
public double[] classifyInstances(Instances instances)
{
double[] ret = new double[instances.numInstances()];
double[] v = distributionForInstances(instances);
for (int i = 0; i < ret.Length; ++i)
{
ret[i] = v[i] > m_delta ? 2 : 0;
}
return ret;
}
//Added by Alain Espinosa///////////////////////////////////////////////////////
/// <summary> Filters an entire set of instances and returnsthe new set.
///
/// </summary>
/// <param name="data">the data to be filtered
/// </param>
/// <returns> the filtered set of data
/// </returns>
/// <exception cref="Exception">if the filter can't be used successfully
/// </exception>
public virtual Instances FilterInstances(Instances data)
{
setInputFormat(data);
for (int i = 0; i < data.numInstances(); i++)
{
input(data.instance(i));
}
batchFinished();
Instances newData = getOutputFormat();
Instance processed;
while ((processed = output()) != null)
{
newData.add(processed);
}
return newData;
}
public void ExecuteBest(DateTime nowDate, List <CandidateClassifier> minScoreInfos)
{
OutputRealDealSummary(nowDate);
float totalCost = 0;
int totalDeal = 0;
if (minScoreInfos != null)
{
List <DealInfo> candidateDeals = new List <DealInfo>();
foreach (var minScoreInfo in minScoreInfos)
{
minScoreInfo.WekaData.GenerateData(false, true);
weka.core.Instances minTestInstances = minScoreInfo.WekaData.CurrentTestInstances;
weka.core.Instances minTestInstancesNew = minScoreInfo.WekaData.CurrentTestInstancesNew;
if (minTestInstances.numInstances() > 0)
{
MyEvaluation eval = new MyEvaluation(minScoreInfo.CostMatrix);
eval.evaluateModel(minScoreInfo.CurrentTestRet, minScoreInfo.CurrentClassValue);
float vol;
// vol = (float)minScoreInfo.MoneyManagement.GetVolume(null);
vol = 0.1F;
//vol = (float)Math.Round(minScore / -20000.0, 1);
//WekaUtils.DebugAssert(vol > 0);
int tp = (int)eval.numTruePositives(1);
int fp = (int)eval.numFalsePositives(1);
double minScore = minScoreInfo.Deals.NowScore;
//WekaUtils.Instance.WriteLog(string.Format("Best Classifier: N={0},TC={1},TP={2},FP={3},TD={4},TV={5},TTP={6},TFP={7},",
// minScoreInfo.Name, minScoreInfo.Deals.NowScore.ToString(Parameters.DoubleFormatString),
// minScoreInfo.Deals.NowTp, minScoreInfo.Deals.NowFp, minScoreInfo.Deals.NowDeal,
// minScoreInfo.Deals.TotalVolume.ToString("N2"), tp, fp),
// true, ConsoleColor.DarkGreen);
// Exclude
//if (TestParameters.EnableExcludeClassifier)
//{
//if (minScoreInfo.ExcludeClassifier == null)
//{
// string modelFileName4Exclude = GetExcludeModelFileName(minScoreInfo.Name);
// minScoreInfo.ExcludeClassifier = WekaUtils.TryLoadClassifier(modelFileName4Exclude);
//}
//}
for (int i = 0; i < minScoreInfo.CurrentTestRet.Length; i++)
{
if (minScoreInfo.CurrentTestRet[i] == 2)
{
if (minScoreInfo.ExcludeClassifier != null)
{
double cv2 = minScoreInfo.ExcludeClassifier.classifyInstance(minTestInstancesNew.instance(i));
if (cv2 != 2)
{
continue;
}
}
candidateDeals.Add(new DealInfo(WekaUtils.GetDateValueFromInstances(minTestInstances, 0, i),
(float)WekaUtils.GetValueFromInstance(minTestInstances, "mainClose", i),
minScoreInfo.DealType,
(float)vol,
(float)(minScoreInfo.CurrentClassValue[i] == 2 ? -minScoreInfo.Tp : minScoreInfo.Sl),
WekaUtils.GetDateValueFromInstances(minTestInstances, 1, i)));
}
}
float nowCost = (float)eval.totalCost();
int nowDeal = tp + fp;
totalCost += nowCost * vol;
totalDeal += nowDeal;
//float diff = Math.Abs(totalCost - realDealsInfo.TotalCost);
//WekaUtils.DebugAssert(diff < 5);
//WekaUtils.DebugAssert(Math.Abs(totalDeal - (realDealsInfo.NowDeal + realDealsInfo.CurrentDeal)) == 0);
//if (diff > 0.5)
// totalCost = realDealsInfo.TotalCost;
IterateClassifierInfos((k, i, j, h) =>
{
if (m_classifierInfoIdxs[k, i, j, h] == minScoreInfo)
{
m_totalScores[k, i, j, h] += nowCost;
m_totalDeals[k, i, j, h] += nowDeal;
return(false);
}
else
{
return(true);
}
});
}
}
int selectCount = 50;
for (int i = 0; i < selectCount; ++i)
{
if (candidateDeals.Count == 0)
{
break;
}
int selectedDealIdx = (int)Math.Round(m_randomGenerator.NextDouble() * candidateDeals.Count);
if (selectedDealIdx == candidateDeals.Count)
{
selectedDealIdx = candidateDeals.Count - 1;
}
m_realDealsInfo.AddDeal(candidateDeals[selectedDealIdx]);
candidateDeals.RemoveAt(selectedDealIdx);
}
}
//if (m_enableDetailLogLevel2)
//{
// IterateClassifierInfos((k, i, j, h) =>
// {
// if (h == m_currentTestHour && m_totalScores[k, i, j, h] != 0)
// {
// WekaUtils.Instance.WriteLog(string.Format("Predict score for {0}: TC={1}, TD={2}", m_classifierInfoIdxs[k, i, j, h].Name, m_totalScores[k, i, j, h].ToString(Parameters.DoubleFormatString), m_totalDeals[k, i, j, h]));
// }
// });
//}
if (!string.IsNullOrEmpty(currentSummary))
{
WekaUtils.Instance.WriteLog(currentSummary, true, ConsoleColor.Red);
System.Console.Title = nowDate.ToString(Parameters.DateTimeFormat) + ":" + currentSummary;
}
//if (m_enableDetailLogLevel2)
//{
// if (TestParameters.EnablePerhourTrain)
// {
// double[] totalCostPerHour = new double[m_classifierInfoIdxs.GetLength(3)];
// int[] totalDealPerHour = new int[m_classifierInfoIdxs.GetLength(3)];
// IterateClassifierInfos((k, i, j, h) =>
// {
// totalCostPerHour[h] += m_totalScores[k, i, j, h];
// totalDealPerHour[h] += m_totalDeals[k, i, j, h];
// });
// for (int i = 0; i < totalCostPerHour.Length; ++i)
// {
// if (totalCostPerHour[i] == 0)
// continue;
// WekaUtils.Instance.WriteLog(string.Format("Predict score Per hour of {0}: TC={1}, TD={2}", i, totalCostPerHour[i].ToString(Parameters.DoubleFormatString), totalDealPerHour[i]));
// }
// }
//if ((m_testTimeStart.Month == 4 || m_testTimeStart.Month == 3) && m_testTimeStart.Day == 20)
//{
// foreach (var cls in m_classifierInfos)
// {
// WekaUtils.Instance.WriteLog(cls.Value.Deals.PrintAll());
// }
//}
//if (m_currentTestHour == 0)
//{
// var cls = m_classifierInfoIdxs[0, 0, 0, 0];
// WekaUtils.Instance.WriteLog((cls.Classifier as RandomClassifier).GetCountInfo());
//}
//}
//if (m_saveDataFile)
//{
// var files = System.IO.Directory.GetFiles(m_baseDir, "*.arff");
// if (files.Length > 100)
// {
// foreach (string fileName in files)
// {
// try
// {
// System.IO.File.Delete(fileName);
// }
// catch (Exception)
// {
// }
// }
// }
//}
}
public override void buildClassifier(Instances instances)
{
m_mustValue = -1;
var weights = GetCount(instances);
if (weights[0] == 0)
{
m_mustValue = 2;
return;
}
else if (weights[2] == 0)
{
m_mustValue = 0;
return;
}
// can classifier handle the data?
getCapabilities().testWithFail(instances);
//instances.deleteWithMissingClass();
Instances trainInstances = null;
if (m_delta == -1)
{
// double delta = 0.5;
// double maxDelta = 1;
// double minDelta = 0.5;
// for (int it = 0; it < 10; ++it)
// {
// delta = (minDelta + maxDelta) / 2.0;
// trainInstances = new Instances(instances, 0, instances.numInstances());
// var cls = TrainOnce(trainInstances, delta);
// double a = 0, b = 0;
// foreach (Instance i in trainInstances)
// {
// var v = cls.classifyInstance(i);
// if (v == 0)
// continue;
// if (i.classValue() == 0)
// a++;
// else if (i.classValue() == 2)
// b++;
// }
// if (b == 0)
// {
// maxDelta = delta;
// }
// else
// {
// minDelta = delta;
// }
// }
// delta = Math.Max(0.5, delta - 0.02);
// double bestDelta = delta;
// double bestCost = double.MaxValue;
// weka.classifiers.Classifier bestCls = null;
// for (int it = 0; it < 20; ++it)
// {
// trainInstances = new Instances(instances, 0, instances.numInstances());
// var cls = TrainOnce(trainInstances, delta);
// MyEvaluation eval = new MyEvaluation(TestParameter.CostMatrix);
// m_Classifier = cls;
// m_delta = delta;
// eval.evaluateModel(this, instances);
// if (eval.totalCost() < bestCost)
// {
// bestDelta = delta;
// bestCost = eval.totalCost();
// bestCls = cls;
// }
// delta += 0.002;
// }
// this.m_Classifier = bestCls;
// this.m_delta = bestDelta;
trainInstances = new Instances(instances, 0, instances.numInstances());
var cls = TrainOnce(trainInstances, 0.5);
this.m_Classifier = cls;
GetBestDelta(instances);
}
else
{
trainInstances = new Instances(instances, 0, instances.numInstances());
var cls = TrainOnce(trainInstances, m_delta);
this.m_Classifier = cls;
}
}
public double[] distributionForInstances(Instances instances)
{
double[] v = new double[instances.numInstances()];
for(int i=0; i<v.Length; ++i)
v[i] = m_Classifier.distributionForInstance(instances.instance(i))[1];
return v;
}
private weka.classifiers.Classifier TrainOnce(Instances trainInstances, double delta)
{
var cls = weka.classifiers.AbstractClassifier.makeCopy(m_Classifier);
bool changed = false;
for (int i = 0; i < trainInstances.numInstances(); ++i)
{
if (trainInstances.instance(i).classValue() == 1)
{
trainInstances.instance(i).setClassMissing();
}
}
for (int i = 0; i < trainInstances.numInstances(); ++i)
{
if (trainInstances.instance(i).classValue() == 2)
{
trainInstances.instance(i).setClassValue(1);
}
}
while (true)
{
changed = false;
double[] w = GetCount(trainInstances);
//string ws = string.Empty;
//if (w != null)
//{
// ws = w[1].ToString(Parameters.DoubleFormatString) + " " + w[0].ToString(Parameters.DoubleFormatString);
//}
double cost = m_tp / m_sl;
string ws;
if (w[1] == 0 || w[0] == 0)
{
ws = string.Empty;
}
else
{
ws = cost.ToString("N2") + " 1";
}
var linear = cls as weka.classifiers.functions.LibLINEAR;
if (linear != null)
{
linear.setWeights(ws);
}
else
{
var svm = cls as weka.classifiers.functions.LibSVM;
if (svm != null)
{
svm.setWeights(ws);
}
}
cls.buildClassifier(trainInstances);
foreach (Instance i in trainInstances)
{
if (i.classValue() == 0)
continue;
double v = cls.classifyInstance(i);
if (v == 0 || cls.distributionForInstance(i)[1] < delta)
{
i.setClassValue(0);
changed = true;
}
if (v == 1)
{
}
}
if (!changed)
break;
}
return cls;
}
/// <summary> Gets the index of the instance with the closest threshold value to the
/// desired target
///
/// </summary>
/// <param name="tcurve">a set of instances that have been generated by this class
/// </param>
/// <param name="threshold">the target threshold
/// </param>
/// <returns> the index of the instance that has threshold closest to
/// the target, or -1 if this could not be found (i.e. no data, or
/// bad threshold target)
/// </returns>
public static int getThresholdInstance(Instances tcurve, double threshold)
{
if (!RELATION_NAME.Equals(tcurve.relationName()) || (tcurve.numInstances() == 0) || (threshold < 0) || (threshold > 1.0))
{
return - 1;
}
if (tcurve.numInstances() == 1)
{
return 0;
}
double[] tvals = tcurve.attributeToDoubleArray(tcurve.numAttributes() - 1);
int[] sorted = Utils.sort(tvals);
return binarySearch(sorted, tvals, threshold);
}
private void button1_Click(object sender, EventArgs e)
{
OpenFileDialog file = new OpenFileDialog();
if (file.ShowDialog() == DialogResult.OK)
{
string filename = file.FileName;
string filee = Path.GetFileName(filename);
bool attributeType;
string attributeName = " ";
int numAttributeValue = 0;
string attributeValueName = " ";
textBox1.Text = filee + " chosen succesfully!";
///////Decision Tree
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader(filename));
insts.setClassIndex(insts.numAttributes() - 1);
//find nominal or numeric attributes and create dropbox or textbox
int numofAttributes = insts.numAttributes() - 1;
for (int i = 0; i < numofAttributes; i++)
{
attributeType = insts.attribute(i).isNumeric();
attributeName = insts.attribute(i).name();
dataGridView1.Rows.Add(attributeName);
if (attributeType == true)
{
}
else
{
numAttributeValue = insts.attribute(i).numValues();
string[] name = new string[numAttributeValue];
for (int j = 0; j < numAttributeValue; j++)
{
attributeValueName = insts.attribute(i).value(j);
name[j] += attributeValueName;
}
DataGridViewComboBoxCell combo = new DataGridViewComboBoxCell();
combo.DataSource = name.ToList();
dataGridView1.Rows[i].Cells[1] = combo;
}
}
cl = new weka.classifiers.trees.J48();
textBox2.Text = "Performing " + percentSplit + "% split evaluation.";
//filling missing values
weka.filters.Filter missingval = new weka.filters.unsupervised.attribute.ReplaceMissingValues();
missingval.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, missingval);
weka.filters.Filter myNormalized = new weka.filters.unsupervised.instance.Normalize();
myNormalized.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, myNormalized);
//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);
string str = cl.toString();
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++;
}
}
textBox3.Text = numCorrect + " out of " + testSize + " correct (" +
(double)((double)numCorrect / (double)testSize * 100.0) + "%)";
//////////Naive Bayes
//dosya okuma
weka.core.Instances insts2 = new weka.core.Instances(new java.io.FileReader(filename));
insts2.setClassIndex(insts2.numAttributes() - 1);
//naive bayes
cl2 = new weka.classifiers.bayes.NaiveBayes();
//filling missing values
weka.filters.Filter missingval2 = new weka.filters.unsupervised.attribute.ReplaceMissingValues();
missingval2.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, missingval2);
//for naive bayes
weka.filters.Filter discrete2 = new weka.filters.unsupervised.attribute.Discretize();
discrete2.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, discrete2);
//randomize the order of the instances in the dataset. -ortak
weka.filters.Filter myRandom2 = new weka.filters.unsupervised.instance.Randomize();
myRandom2.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myRandom2);
//ortak
int trainSize2 = insts2.numInstances() * percentSplit / 100;
int testSize2 = insts2.numInstances() - trainSize2;
weka.core.Instances train2 = new weka.core.Instances(insts2, 0, trainSize2);
cl2.buildClassifier(train2);
string str2 = cl2.toString();
int numCorrect2 = 0;
for (int i = trainSize2; i < insts2.numInstances(); i++)
{
weka.core.Instance currentInst2 = insts2.instance(i);
double predictedClass2 = cl2.classifyInstance(currentInst2);
if (predictedClass2 == insts2.instance(i).classValue())
{
numCorrect2++;
}
}
textBox4.Text = numCorrect2 + " out of " + testSize2 + " correct (" +
(double)((double)numCorrect2 / (double)testSize2 * 100.0) + "%)";
/////////K-Nearest Neigbour
//dosya okuma
weka.core.Instances insts3 = new weka.core.Instances(new java.io.FileReader(filename));
insts3.setClassIndex(insts3.numAttributes() - 1);
cl3 = new weka.classifiers.lazy.IBk();
//filling missing values
weka.filters.Filter missingval3 = new weka.filters.unsupervised.attribute.ReplaceMissingValues();
missingval3.setInputFormat(insts3);
insts3 = weka.filters.Filter.useFilter(insts3, missingval3);
//Convert to dummy attribute knn,svm,neural network
weka.filters.Filter dummy3 = new weka.filters.unsupervised.attribute.NominalToBinary();
dummy3.setInputFormat(insts3);
insts3 = weka.filters.Filter.useFilter(insts3, dummy3);
//normalize numeric attribute
weka.filters.Filter myNormalized3 = new weka.filters.unsupervised.instance.Normalize();
myNormalized3.setInputFormat(insts3);
insts3 = weka.filters.Filter.useFilter(insts3, myNormalized3);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom3 = new weka.filters.unsupervised.instance.Randomize();
myRandom3.setInputFormat(insts3);
insts3 = weka.filters.Filter.useFilter(insts3, myRandom3);
int trainSize3 = insts3.numInstances() * percentSplit / 100;
int testSize3 = insts3.numInstances() - trainSize3;
weka.core.Instances train3 = new weka.core.Instances(insts3, 0, trainSize3);
cl3.buildClassifier(train3);
string str3 = cl3.toString();
int numCorrect3 = 0;
for (int i = trainSize3; i < insts3.numInstances(); i++)
{
weka.core.Instance currentInst3 = insts3.instance(i);
double predictedClass3 = cl3.classifyInstance(currentInst3);
if (predictedClass3 == insts3.instance(i).classValue())
{
numCorrect3++;
}
}
textBox5.Text = numCorrect3 + " out of " + testSize3 + " correct (" +
(double)((double)numCorrect3 / (double)testSize3 * 100.0) + "%)";
//////////Artificial neural network
//dosya okuma
weka.core.Instances insts4 = new weka.core.Instances(new java.io.FileReader(filename));
insts4.setClassIndex(insts4.numAttributes() - 1);
cl4 = new weka.classifiers.functions.MultilayerPerceptron();
//filling missing values
weka.filters.Filter missingval4 = new weka.filters.unsupervised.attribute.ReplaceMissingValues();
missingval4.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, missingval4);
//Convert to dummy attribute
weka.filters.Filter dummy4 = new weka.filters.unsupervised.attribute.NominalToBinary();
dummy4.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, dummy4);
//normalize numeric attribute
weka.filters.Filter myNormalized4 = new weka.filters.unsupervised.instance.Normalize();
myNormalized4.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, myNormalized4);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom4 = new weka.filters.unsupervised.instance.Randomize();
myRandom4.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, myRandom4);
int trainSize4 = insts4.numInstances() * percentSplit / 100;
int testSize4 = insts4.numInstances() - trainSize4;
weka.core.Instances train4 = new weka.core.Instances(insts4, 0, trainSize4);
cl4.buildClassifier(train4);
string str4 = cl4.toString();
int numCorrect4 = 0;
for (int i = trainSize4; i < insts4.numInstances(); i++)
{
weka.core.Instance currentInst4 = insts4.instance(i);
double predictedClass4 = cl4.classifyInstance(currentInst4);
if (predictedClass4 == insts4.instance(i).classValue())
{
numCorrect4++;
}
}
textBox6.Text = numCorrect4 + " out of " + testSize4 + " correct (" +
(double)((double)numCorrect4 / (double)testSize4 * 100.0) + "%)";
///////Support Vector Machine
// dosya okuma
weka.core.Instances insts5 = new weka.core.Instances(new java.io.FileReader(filename));
insts5.setClassIndex(insts5.numAttributes() - 1);
cl5 = new weka.classifiers.functions.SMO();
//filling missing values
weka.filters.Filter missingval5 = new weka.filters.unsupervised.attribute.ReplaceMissingValues();
missingval5.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, missingval5);
//Convert to dummy attribute
weka.filters.Filter dummy5 = new weka.filters.unsupervised.attribute.NominalToBinary();
dummy5.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, dummy5);
//normalize numeric attribute
weka.filters.Filter myNormalized5 = new weka.filters.unsupervised.instance.Normalize();
myNormalized5.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, myNormalized5);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom5 = new weka.filters.unsupervised.instance.Randomize();
myRandom5.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, myRandom5);
int trainSize5 = insts5.numInstances() * percentSplit / 100;
int testSize5 = insts5.numInstances() - trainSize5;
weka.core.Instances train5 = new weka.core.Instances(insts5, 0, trainSize5);
cl5.buildClassifier(train5);
string str5 = cl5.toString();
int numCorrect5 = 0;
for (int i = trainSize5; i < insts5.numInstances(); i++)
{
weka.core.Instance currentInst5 = insts5.instance(i);
double predictedClass5 = cl5.classifyInstance(currentInst5);
if (predictedClass5 == insts5.instance(i).classValue())
{
numCorrect5++;
}
}
textBox7.Text = numCorrect5 + " out of " + testSize5 + " correct (" +
(double)((double)numCorrect5 / (double)testSize5 * 100.0) + "%)";
string result1 = textBox3.Text;
string output1 = result1.Split('(', ')')[1];
output1 = output1.Remove(output1.Length - 1);
double r1 = Convert.ToDouble(output1);
string result2 = textBox4.Text;
string output2 = result2.Split('(', ')')[1];
output2 = output2.Remove(output2.Length - 1);
double r2 = Convert.ToDouble(output2);
string result3 = textBox5.Text;
string output3 = result3.Split('(', ')')[1];
output3 = output3.Remove(output3.Length - 1);
double r3 = Convert.ToDouble(output3);
string result4 = textBox6.Text;
string output4 = result4.Split('(', ')')[1];
output4 = output4.Remove(output4.Length - 1);
double r4 = Convert.ToDouble(output4);
string result5 = textBox7.Text;
string output5 = result5.Split('(', ')')[1];
output5 = output5.Remove(output5.Length - 1);
double r5 = Convert.ToDouble(output5);
double[] max_array = new double[] { r1, r2, r3, r4, r5 };
double max = max_array.Max();
if (r1 == max)
{
textBox8.Text = "Best Algoritm is Decision Tree Algorithm ";
}
else if (r2 == max)
{
textBox8.Text = "Best Algoritm is Naive Bayes Algorithm ";
}
else if (r3 == max)
{
textBox8.Text = "Best Algoritm is K-Nearest Neighbour Algorithm ";
}
else if (r4 == max)
{
textBox8.Text = "Best Algoritm is Artificial Neural Network Algorithm ";
}
else if (r5 == max)
{
textBox8.Text = "Best Algoritm is Support Vector Machine Algorithm ";
}
}
}
public static string classifyTest(string file, string classifier)
{
string data = "No data";
try
{
//weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader("C:\\Users\\kinli\\source\\repos\\WebApplication2\\WebApplication2\\iris.arff"));
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader(file));
insts.setClassIndex(insts.numAttributes() - 1);
weka.classifiers.Classifier cl = new weka.classifiers.trees.J48();
if (classifier == "J48")
{
cl = new weka.classifiers.trees.J48();
}
else if (classifier == "MLP")
{
cl = new weka.classifiers.functions.MultilayerPerceptron();
}
else if (classifier == "NaiveBayes")
{
cl = new weka.classifiers.bayes.NaiveBayes();
}
//data = ("Performing " + percentSplit + "% split evaluation.\n");
data = ("Performing use training set evaluation.\n");
//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++;
* }*/
cl.buildClassifier(insts);
int numCorrect = 0;
for (int i = 0; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = cl.classifyInstance(currentInst);
if (predictedClass == insts.instance(i).classValue())
{
numCorrect++;
}
}
data = data + (numCorrect + " out of " + insts.numInstances() + " correct (" +
(double)((double)numCorrect / (double)insts.numInstances() * 100.0) + "%)");
}
catch (java.lang.Exception ex)
{
data = "Error";
ex.printStackTrace();
}
return(data);
}
private void Classify(string path)
{
readyToTest = false; // initialize flag
// Try reading file, if failed exit function
insts = ReadFile(path);
if (insts == null)
{
// Error occured reading file, display error message
MessageBox.Show("Instances are null!", "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
return;
}
var form = Form.ActiveForm as Form1; // get the current form object
// Reset UI and lists
succesRates.Clear();
classifiers.Clear();
form.inputPanel.Controls.Clear();
inputObjects.Clear();
form.textMostSuccessful.Text = "";
form.testResult.Text = "";
// Place attribute inputs on UI, max 18, numeric and nominal
int offsetV = 60;
int offsetH = 10;
int width = 75;
int height = 30;
for (int i = 0; i < insts.numAttributes() - 1; i++)
{
// Create and place label
Label label = new Label();
label.Width = width;
label.Height = height;
label.Text = insts.attribute(i).name();
label.Parent = form.inputPanel;
label.Location = new Point((width * (i % 8)) + offsetH, (height * (i / 8)) + (offsetV * (i / 8)));
// NumericUpDown for numeric and ComboBox for nominal values
if (insts.attribute(i).isNumeric())
{
NumericUpDown nud = new NumericUpDown();
nud.Width = width - 10;
nud.Height = height;
nud.Parent = form.inputPanel;
nud.Location = new Point((width * (i % 8)) + offsetH, (height * (i / 8)) + (offsetV * (i / 8)) + height);
inputObjects.Add(new UserInput(nud));
}
else
{
string[] values = insts.attribute(i).toString().Split('{', '}')[1].Split(',');
ComboBox comboBox = new ComboBox();
comboBox.DataSource = values;
comboBox.Width = width - 10;
comboBox.Height = height;
comboBox.Parent = form.inputPanel;
comboBox.Location = new Point((width * (i % 8)) + offsetH, (height * (i / 8)) + (offsetV * (i / 8)) + height);
inputObjects.Add(new UserInput(comboBox));
}
}
// Set train and test sizes
trainSize = insts.numInstances() * percentSplit / 100;
testSize = insts.numInstances() - trainSize;
// Set target attribute
insts.setClassIndex(insts.numAttributes() - 1);
// Randomize
weka.filters.Filter rndFilter = new weka.filters.unsupervised.instance.Randomize();
rndFilter.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, rndFilter);
// Start threads for each method
Thread t_SuccessNb = new Thread(() => CalculateSuccessForNb(insts));
t_SuccessNb.Start();
Thread t_SuccessKn = new Thread(() => CalculateSuccessForKn(insts));
t_SuccessKn.Start();
Thread t_SuccessDt = new Thread(() => CalculateSuccessForDt(insts));
t_SuccessDt.Start();
Thread t_SuccessAnn = new Thread(() => CalculateSuccessForAnn(insts));
t_SuccessAnn.Start();
Thread t_SuccessSvm = new Thread(() => CalculateSuccessForSvm(insts));
t_SuccessSvm.Start();
// Wait for threads
t_SuccessNb.Join();
t_SuccessKn.Join();
t_SuccessDt.Join();
t_SuccessAnn.Join();
t_SuccessSvm.Join();
// Find out which algorithm has the best success rate
foreach (var item in succesRates)
{
if (highestSuccessRate.Equals(default(KeyValuePair <Classifier, double>)) || highestSuccessRate.Value < item.Value)
{
highestSuccessRate = item;
}
}
form.textMostSuccessful.Text = "Most successful algorithm is " + highestSuccessRate.Key + " and it will be used for testing.";
readyToTest = true; // switch flag
}
/// <summary> Calculates the n point precision result, which is the precision averaged
/// over n evenly spaced (w.r.t recall) samples of the curve.
///
/// </summary>
/// <param name="tcurve">a previously extracted threshold curve Instances.
/// </param>
/// <param name="n">the number of points to average over.
/// </param>
/// <returns> the n-point precision.
/// </returns>
public static double getNPointPrecision(Instances tcurve, int n)
{
if (!RELATION_NAME.Equals(tcurve.relationName()) || (tcurve.numInstances() == 0))
{
return System.Double.NaN;
}
int recallInd = tcurve.attribute(RECALL_NAME).index();
int precisInd = tcurve.attribute(PRECISION_NAME).index();
double[] recallVals = tcurve.attributeToDoubleArray(recallInd);
int[] sorted = Utils.sort(recallVals);
double isize = 1.0 / (n - 1);
double psum = 0;
for (int i = 0; i < n; i++)
{
int pos = binarySearch(sorted, recallVals, i * isize);
double recall = recallVals[sorted[pos]];
double precis = tcurve.instance(sorted[pos]).value_Renamed(precisInd);
/*
System.err.println("Point " + (i + 1) + ": i=" + pos
+ " r=" + (i * isize)
+ " p'=" + precis
+ " r'=" + recall);
*/
// interpolate figures for non-endpoints
while ((pos != 0) && (pos < sorted.Length - 1))
{
pos++;
double recall2 = recallVals[sorted[pos]];
if (recall2 != recall)
{
double precis2 = tcurve.instance(sorted[pos]).value_Renamed(precisInd);
double slope = (precis2 - precis) / (recall2 - recall);
double offset = precis - recall * slope;
precis = isize * i * slope + offset;
/*
System.err.println("Point2 " + (i + 1) + ": i=" + pos
+ " r=" + (i * isize)
+ " p'=" + precis2
+ " r'=" + recall2
+ " p''=" + precis);
*/
break;
}
}
psum += precis;
}
return psum / n;
}
public static string IncrementTest(weka.core.Instances allInstancesWithDate,
Func<weka.classifiers.Classifier> clsCreator, string removeAttributes, string resultFile, int step)
{
//if (!(TestParameters2.UsePartialHpDataM1 || TestParameters2.UsePartialHpData))
//{
// HpData.Instance.Clear();
//}
int trainMinutes = TestParameters2.MinTrainPeriod * WekaUtils.GetMinuteofPeriod(TestParameters2.CandidateParameter.MainPeriod);
string ret = string.Empty;
string sampleFile = null; // resultFile.Replace("Increment", "sample");
bool useInstanceWeight = false;
bool enablePerHour = false;
bool enableDiffClass = false;
int sameClassCount = -1;// TestParameters2.MaxTrainSize / 3; // allInstances.numClasses();
bool enableDiffHpTime = false;
bool enableRemoveLittle = false;
bool enableRemoveLargeThanMid = true;
bool enableFilter = false;
if (!TestParameters2.RealTimeMode && File.Exists(resultFile))
File.Delete(resultFile);
weka.core.Instances allInstances;
var filter = new weka.filters.MultiFilter();
//filter.setOptions(weka.core.Utils.splitOptions("-F \"weka.filters.unsupervised.attribute.Remove -R 1,4\" -F \"weka.filters.unsupervised.attribute.Discretize -B 10 -M -1.0 -R first-last\""));
filter.setOptions(weka.core.Utils.splitOptions(string.Format("-F \"weka.filters.unsupervised.attribute.Remove -R {0} \"", removeAttributes)));
filter.setInputFormat(allInstancesWithDate);
allInstances = weka.filters.Filter.useFilter(allInstancesWithDate, filter);
long[] jHpTimes = new long[allInstancesWithDate.numInstances()];
DateTime[] jDates = new DateTime[allInstancesWithDate.numInstances()];
DateTime[] jHpDates = new DateTime[allInstancesWithDate.numInstances()];
int[] jHps = new int[allInstancesWithDate.numInstances()];
for (int j = 0; j < jDates.Length; ++j)
{
jDates[j] = WekaUtils.GetDateValueFromInstances(allInstancesWithDate, 0, j);
jHpTimes[j] = WekaUtils.GetTimeValueFromInstances(allInstancesWithDate, 1, j);
jHpDates[j] = WekaUtils.GetDateFromTime(jHpTimes[j]);
jHps[j] = (int)allInstances.instance(j).classValue();
}
#region "action"
Func<Tuple<int, int>, Tuple<weka.core.Instances, weka.core.Instances, long, int, int>> action = (tuple) =>
{
var ai = tuple.Item1;
DateTime nowDate = WekaUtils.GetDateValueFromInstances(allInstancesWithDate, 0, ai);
if (nowDate < TestParameters2.TrainStartTime || nowDate > TestParameters2.TrainEndTime)
return null;
DateTime nowHpDate = WekaUtils.GetDateValueFromInstances(allInstancesWithDate, 1, ai);
double nowClass = allInstancesWithDate.instance(ai).classValue();
double preJClass = -1;
List<weka.core.Instance> listTrainInstances = new List<weka.core.Instance>(ai / 2);
List<weka.core.Instance> listTrainInstancesWithDate = new List<weka.core.Instance>(ai / 2);
int[] counts = new int[allInstancesWithDate.numClasses()];
long maxjHpTime = -1;
int maxjHpTimeHp = 2;
int maxjHpTimeCount = 0;
bool enoughTrainMinutes = false;
DateTime firstDate = nowDate.AddMinutes(-trainMinutes);
int classIdxWithDate = allInstancesWithDate.classIndex();
int classIdx = allInstances.classIndex();
for (int j = ai - 1; j >= 0; --j)
{
long jHpTime = jHpTimes[j]; // WekaUtils.GetTimeValueFromInstances(allInstancesWithDate, 1, j);
DateTime jHpDate = jHpDates[j];// WekaUtils.GetDateFromTime(jHpTime);
DateTime jDate = jDates[j];// WekaUtils.GetDateValueFromInstances(allInstancesWithDate, 0, j);
int jHp = jHps[j]; // (int)allInstancesWithDate.instance(j).value(classIdxWithDate);
if (enablePerHour)
{
if (nowDate.Hour != jDate.Hour)
continue;
}
weka.core.Instance instInsert = null;
weka.core.Instance instInsertWithDate = null;
if (jHpDate <= nowDate)
{
if (jHpTime > maxjHpTime)
{
maxjHpTime = jHpTime;
maxjHpTimeHp = jHp;
maxjHpTimeCount = 1;
}
else if (jHpTime == maxjHpTime)
{
maxjHpTimeCount++;
}
instInsert = new weka.core.DenseInstance(allInstances.instance(j));
//instInsert.setDataset(trainInstances);
instInsertWithDate = new weka.core.DenseInstance(allInstancesWithDate.instance(j));
//instInsertWithDate.setDataset(trainInstancesWithDate);
}
else
{
if (TestParameters2.UsePartialHpDataM1 || TestParameters2.UsePartialHpData)
{
Tuple<int, long> hp = null;
if (TestParameters2.UsePartialHpDataM1)
{
hp = HpData.Instance.GetHpSumByM1(TestParameters2.CandidateParameter.MainSymbol, TestParameters2.CandidateParameter.MainPeriod,
WekaUtils.GetTimeFromDate(nowDate), WekaUtils.GetTimeFromDate(jDate));
if (hp.Item2 == 0)
hp = null;
}
else if (TestParameters2.UsePartialHpData)
{
var hps = HpData.Instance.GetHpSum(TestParameters2.CandidateParameter.MainSymbol, TestParameters2.CandidateParameter.MainPeriod,
WekaUtils.GetTimeFromDate(nowDate), WekaUtils.GetTimeFromDate(jDate));
if (hps.ContainsKey(jDate))
{
hp = hps[jDate];
}
}
if (hp != null)
{
if (WekaUtils.GetDateFromTime(hp.Item2) > nowDate)
{
throw new AssertException("hpdate should less than now");
}
jHp = hp.Item1;
jHpTime = hp.Item2;
if (jHpTime > maxjHpTime)
{
maxjHpTime = jHpTime;
maxjHpTimeHp = jHp;
maxjHpTimeCount = 0;
}
else if (jHpTime == maxjHpTime)
{
maxjHpTimeCount++;
}
instInsert = new weka.core.DenseInstance(allInstances.instance(j));
//instInsert.setDataset(trainInstances);
//instInsert.setClassValue(jHp);
instInsert.setValue(classIdx, jHp);
instInsertWithDate = new weka.core.DenseInstance(allInstancesWithDate.instance(j));
//instInsertWithDate.setDataset(trainInstancesWithDate);
//instInsertWithDate.setClassValue(jHp);
instInsertWithDate.setValue(classIdxWithDate, jHp);
instInsertWithDate.setValue(1, jHpTime * 1000);
}
}
}
if (instInsert == null)
continue;
double jClass = jHp;
if (enableDiffClass && jClass == preJClass)
continue;
if (sameClassCount > 0)
{
if (counts[(int)jClass] >= sameClassCount)
continue;
counts[(int)jClass]++;
}
if (enableFilter && j > 0 && Filter(jDate, allInstancesWithDate.instance(j), allInstancesWithDate.instance(j - 1)))
continue;
if (useInstanceWeight)
instInsert.setWeight((nowDate - jDate).TotalMinutes);
listTrainInstances.Add(instInsert);
listTrainInstancesWithDate.Add(instInsertWithDate);
preJClass = jClass;
if (jDate <= firstDate)
{
enoughTrainMinutes = true;
break;
}
}
//weka.core.Instances trainInstances2 = new weka.core.Instances(allInstancesNoDate, 0);
//for (int x = trainInstances.numInstances() - 1; x >= 0; --x)
//{
// weka.core.Instance inst = new weka.core.DenseInstance(trainInstances.instance(x));
// trainInstances2.add(inst);
//}
//WekaUtils.SaveInstances(trainInstances2, "d:\\a.arff");
//if (trainInstances.numInstances() >= trainLength)
// break;
if (!enoughTrainMinutes)
{
Console.WriteLine(string.Format("{0}, not enough trainMinutes",
nowDate.ToString(Parameters.DateTimeFormat)));
return null;
}
if (listTrainInstances.Count < TestParameters2.MinTrainSize)
{
Console.WriteLine(string.Format("{0}, numInstances {1} < minTrainSize {2}",
nowDate.ToString(Parameters.DateTimeFormat), listTrainInstances.Count, TestParameters2.MinTrainSize));
return null;
}
//else if (listTrainInstances.Count == 1)
//{
// lock (WekaUtils.Instance)
// {
// WekaUtils.Instance.WriteLog("trainInstances.numInstances() == 1, nowDate = " + nowDate.ToString());
// if (!System.IO.File.Exists("d:\\a.arff"))
// {
// WekaUtils.SaveInstances(trainInstances, "d:\\a.arff");
// }
// }
//}
weka.core.Instances trainInstances = new weka.core.Instances(allInstances, listTrainInstances.Count);
weka.core.Instances trainInstancesWithDate = new weka.core.Instances(allInstancesWithDate, listTrainInstancesWithDate.Count);
WekaUtils.AddInstanceQuickly(trainInstances, listTrainInstances);
WekaUtils.AddInstanceQuickly(trainInstancesWithDate, listTrainInstancesWithDate);
if (enableRemoveLittle)
{
double preClass = 2;
for (int ii = 0; ii < trainInstances.numInstances(); ++ii)
{
var iiClass = trainInstances.instance(ii).classValue();
if (iiClass == 2)
continue;
int jj = ii + 1;
while (jj < trainInstances.numInstances())
{
if (trainInstances.instance(jj).classValue() == iiClass)
jj++;
else
break;
}
int count = jj - ii;
if (count < 5)
{
for (jj = 0; jj < count; ++jj)
{
trainInstances.instance(ii + jj).setClassValue(preClass);
}
}
else
{
preClass = iiClass;
ii += count;
}
}
}
if (enableDiffHpTime)
{
Dictionary<long, int> jDictHpTimes = new Dictionary<long, int>();
int n = trainInstances.numInstances();
List<weka.core.Instance> list1 = new List<weka.core.Instance>(n);
List<weka.core.Instance> list2 = new List<weka.core.Instance>(n);
//java.util.LinkedList deleteList = new java.util.LinkedList();
for (int j = 0; j < n; ++j)
{
long jHpTime = WekaUtils.GetTimeValueFromInstances(trainInstancesWithDate, 1, j);
if (jDictHpTimes.ContainsKey(jHpTime))
{
continue;
}
else
{
jDictHpTimes[jHpTime] = list1.Count;
list1.Add(trainInstances.instance(j));
list2.Add(trainInstancesWithDate.instance(j));
}
}
weka.core.Instances newTrainInstances = new weka.core.Instances(trainInstances, list1.Count);
weka.core.Instances newTrainInstancesWithDate = new weka.core.Instances(trainInstancesWithDate, list2.Count);
WekaUtils.AddInstanceQuickly(newTrainInstances, list1);
WekaUtils.AddInstanceQuickly(newTrainInstancesWithDate, list2);
trainInstances = newTrainInstances;
trainInstancesWithDate = newTrainInstancesWithDate;
}
if (enableRemoveLargeThanMid)
{
int n = trainInstances.numInstances();
long[] lastTimes = new long[n];
for (int j = 0; j < n; ++j)
{
long openTime = WekaUtils.GetTimeValueFromInstances(trainInstancesWithDate, 0, j);
long closeTime = WekaUtils.GetTimeValueFromInstances(trainInstancesWithDate, 1, j);
lastTimes[j] = (long)(closeTime - openTime);
}
Array.Sort(lastTimes);
long midLastTime = lastTimes[lastTimes.Count() / 2];
List<weka.core.Instance> list1 = new List<weka.core.Instance>(n);
List<weka.core.Instance> list2 = new List<weka.core.Instance>(n);
//java.util.LinkedList deleteList = new java.util.LinkedList();
for (int j = 0; j < n; ++j)
{
if (lastTimes[j] > midLastTime)
{
//deleteList.add(trainInstances.instance(j));
}
else
{
list1.Add(trainInstances.instance(j));
list2.Add(trainInstancesWithDate.instance(j));
}
}
weka.core.Instances newTrainInstances = new weka.core.Instances(trainInstances, list1.Count);
weka.core.Instances newTrainInstancesWithDate = new weka.core.Instances(trainInstancesWithDate, list2.Count);
WekaUtils.AddInstanceQuickly(newTrainInstances, list1);
WekaUtils.AddInstanceQuickly(newTrainInstancesWithDate, list2);
trainInstances = newTrainInstances;
trainInstancesWithDate = newTrainInstancesWithDate;
//trainInstances.removeAll(deleteList);
}
if (!string.IsNullOrEmpty(sampleFile))
{
lock (sampleFile)
{
if (!System.IO.File.Exists(sampleFile))
{
WekaUtils.SaveInstances(trainInstancesWithDate, sampleFile);
}
}
}
//using (StreamWriter sw = new StreamWriter("d:\\p.txt", true))
//{
// sw.Write("{0},{1},", nowDate.ToString(Parameters.DateTimeFormat), nowHpDate.ToString(Parameters.DateTimeFormat));
//}
return new Tuple<weka.core.Instances, weka.core.Instances, long, int, int>(trainInstances, trainInstancesWithDate, maxjHpTime, maxjHpTimeHp, maxjHpTimeCount);
};
#endregion
//allInstancesNoDate = allInstances;
if (!TestParameters2.RealTimeMode)
{
int tpb = 0, fpb = 0, tps = 0, fps = 0;
int db = 0, ds = 0, dn = 0;
int parallelStep = 1;
if (TestParameters.EnableMultiThread)
{
parallelStep = 100;
}
int startIdx = TestParameters2.MinTrainPeriod * 2 / 3;
for (int i0 = startIdx; i0 < allInstancesWithDate.numInstances() - step; i0 += step * parallelStep)
{
List<Tuple<int, int>> toTest = new List<Tuple<int, int>>();
for (int i = i0; i < Math.Min(i0 + step * parallelStep, allInstancesWithDate.numInstances() - step); i += step)
{
toTest.Add(new Tuple<int, int>(i, toTest.Count));
}
var toRet0 = new Tuple<weka.core.Instances, weka.core.Instances, long, int, int>[toTest.Count];
if (TestParameters.EnableMultiThread)
{
Parallel.ForEach(toTest, (tuple) =>
{
int i = tuple.Item1 - i0;
toRet0[i] = action(tuple);
});
}
else
{
for (int i = 0; i < toTest.Count; ++i)
{
var r = action(toTest[i]);
toRet0[i] = r;
}
}
object[, ,] toRet = new object[toTest.Count, step, 9];
bool[] toRet1 = new bool[toTest.Count];
for (int i = 0; i < toTest.Count; ++i)
{
for (int j = 0; j < step; ++j)
toRet[i, j, 0] = 2;
if (toRet0[i] != null)
{
var maxjHpTime = toRet0[i].Item3;
var maxjHpTimeHp = toRet0[i].Item4;
var maxjHpTimeCount = toRet0[i].Item5;
toRet1[i] = IncrementTestTrainCheck(resultFile, maxjHpTime, maxjHpTimeHp, maxjHpTimeCount, toTest[i]);
}
}
if (TestParameters.EnableMultiThread)
{
Parallel.ForEach(toTest, (tuple) =>
{
int i = tuple.Item1 - i0;
if (toRet0[i] != null)
{
var trainInstances = toRet0[i].Item1;
var trainInstancesWithDate = toRet0[i].Item2;
var maxjHpTime = toRet0[i].Item3;
var maxjHpTimeHp = toRet0[i].Item4;
var r = IncrementTestTrain(toRet1[i], clsCreator, resultFile, step, trainInstances, trainInstancesWithDate, allInstances, allInstancesWithDate,
maxjHpTime, maxjHpTimeHp, toTest[i]);
if (r != null)
{
for (int j = 0; j < toRet.GetLength(1); ++j)
for (int k = 0; k < toRet.GetLength(2); ++k)
toRet[i, j, k] = r[j, k];
}
}
});
}
else
{
for (int i = 0; i < toTest.Count; ++i)
{
if (toRet0[i] != null)
{
var trainInstances = toRet0[i].Item1;
var trainInstancesWithDate = toRet0[i].Item2;
var maxjHpTime = toRet0[i].Item3;
var maxjHpTimeHp = toRet0[i].Item4;
var r = IncrementTestTrain(toRet1[i], clsCreator, resultFile, step, trainInstances, trainInstancesWithDate, allInstances, allInstancesWithDate,
maxjHpTime, maxjHpTimeHp, toTest[i]);
if (r != null)
{
for (int j = 0; j < toRet.GetLength(1); ++j)
for (int k = 0; k < toRet.GetLength(2); ++k)
toRet[i, j, k] = r[j, k];
}
}
}
}
for (int i = 0; i < toTest.Count; ++i)
{
//action(toTest[i]);
for (int j = 0; j < step; ++j)
{
int d = (int)toRet[i, j, 0];
if (d != 0 && d != 1 && d != 2)
throw new AssertException("d should be -1, 0, 1 or 2. but it's" + d.ToString());
if (toRet[i, j, 6] == null)
{
//Console.WriteLine("toRet[i, j, 6] == null");
continue;
}
long maxJHpTime = (long)toRet[i, j, 6];
//if (lastMaxJHpTime >= (long)toRet[i, j, 6])
// continue;
//lastMaxJHpTime = (long)toRet[i, j, 6];
if (d == 0)
db++;
else if (d == 1)
ds++;
else
dn++;
int v = (int)toRet[i, j, 1];
if (d != 2)
{
if (v == 3 || d == v)
{
if (d == 0)
{
tpb++;
}
else if (d == 1)
{
tps++;
}
}
else
{
if (d == 0)
{
fpb++;
}
else if (d == 1)
{
fps++;
}
}
}
DateTime nowDate = WekaUtils.GetDateFromTime((long)toRet[i, j, 3]);
DateTime nextDate = WekaUtils.GetDateFromTime((long)toRet[i, j, 2]);
//nextDate = WekaUtils.GetDateFromTime(maxJHpTime);
//if (j == 0)
{
if (tpb + fpb + tps + fps != 0)
{
ret = (string.Format("{0}, tn={1},tc={2}, db={3},ds={4},dn={5}, tpb={6},fpb={7},tps={8},fps={9},p={10}",
nextDate.ToString(Parameters.DateTimeFormat), toRet[i, j, 4], toRet[i, j, 5],
db, ds, dn,
tpb, fpb, tps, fps, ((double)(tpb+tps) / (tpb + fpb + tps + fps)).ToString("F2")));
}
else
{
ret = (string.Format("{0}, tn={1},tc={2}, db={3},ds={4},dn={5}, tpb={6},fpb={7},tps={8},fps={9},p={10}",
nextDate.ToString(Parameters.DateTimeFormat), toRet[i, j, 4], toRet[i, j, 5],
db, ds, dn,
tpb, fpb, tps, fps, 0));
}
WekaUtils.Instance.WriteLog(ret);
}
//if (d != 2)
{
using (StreamWriter sw = new StreamWriter(resultFile, true))
{
sw.WriteLine(string.Format("{1}, {2}, {6}, {3}, {4}, {5}", nowDate,
nextDate.ToString(Parameters.DateTimeFormat),
d, v, (double)toRet[i, j, 7],
WekaUtils.GetDateFromTime(maxJHpTime).ToString(Parameters.DateTimeFormat),
toRet[i, j, 8]));
}
}
}
}
}
}
else
{
for (int i0 = allInstancesWithDate.numInstances() - 1; i0 < allInstancesWithDate.numInstances() && i0 >= 0; i0++)
{
List<Tuple<int, int>> toTest = new List<Tuple<int, int>>();
toTest.Add(new Tuple<int, int>(i0, toTest.Count));
var toRet0 = new Tuple<weka.core.Instances, weka.core.Instances, long, int, int>[toTest.Count];
for (int i = 0; i < toTest.Count; ++i)
{
var r = action(toTest[i]);
toRet0[i] = r;
}
object[, ,] toRet = new object[toTest.Count, step, 9];
for (int i = 0; i < toTest.Count; ++i)
{
for (int j = 0; j < toRet.GetLength(1); ++j)
toRet[i, j, 0] = 2;
if (toRet0[i] != null)
{
var trainInstances = toRet0[i].Item1;
var trainInstancesWithDate = toRet0[i].Item2;
var maxjHpTime = toRet0[i].Item3;
var maxjHpTimeHp = toRet0[i].Item4;
var maxjHpTimeCount = toRet0[i].Item5;
bool b = IncrementTestTrainCheck(resultFile, maxjHpTime, maxjHpTimeHp, maxjHpTimeCount, toTest[i]);
var r = IncrementTestTrain(b, clsCreator, resultFile, step, trainInstances, trainInstancesWithDate, allInstances, allInstancesWithDate,
maxjHpTime, maxjHpTimeHp, toTest[i]);
if (r != null)
{
for (int j = 0; j < toRet.GetLength(1); ++j)
for (int k = 0; k < toRet.GetLength(2); ++k)
toRet[i, j, k] = r[j, k];
}
}
}
for (int j = 0; j < step; ++j)
{
int d = (int)toRet[0, j, 0];
if (d != 0 && d != 1 && d != 2)
throw new AssertException("d should be -1, 0, 1 or 2.");
if (toRet[0, j, 6] == null)
continue;
long maxJHpTime = (long)toRet[0, j, 6];
//if (lastMaxJHpTime >= (long)toRet[i, j, 6])
// continue;
//lastMaxJHpTime = (long)toRet[i, j, 6];
int v = (int)toRet[0, j, 1];
DateTime nowDate = WekaUtils.GetDateFromTime((long)toRet[0, j, 3]);
DateTime nextDate = WekaUtils.GetDateFromTime((long)toRet[0, j, 2]);
//nextDate = WekaUtils.GetDateFromTime(maxJHpTime);
//if (d != 2)
{
using (StreamWriter sw = new StreamWriter(resultFile, true))
{
sw.WriteLine(string.Format("{1}, {2}, {6}, {3}, {4}, {5}", nowDate,
nextDate.ToString(Parameters.DateTimeFormat),
d, v, (double)toRet[0, j, 7],
WekaUtils.GetDateFromTime(maxJHpTime).ToString(Parameters.DateTimeFormat),
toRet[0, j, 8]));
}
}
}
}
}
return ret;
}
/// <summary> Applies the cost matrix to a set of instances. If a random number generator is
/// supplied the instances will be resampled, otherwise they will be rewighted.
/// Adapted from code once sitting in Instances.java
///
/// </summary>
/// <param name="data">the instances to reweight.
/// </param>
/// <param name="random">a random number generator for resampling, if null then instances are
/// rewighted.
/// </param>
/// <returns> a new dataset reflecting the cost of misclassification.
/// </returns>
/// <exception cref="Exception">if the data has no class or the matrix in inappropriate.
/// </exception>
public virtual Instances applyCostMatrix(Instances data, System.Random random)
{
double sumOfWeightFactors = 0, sumOfMissClassWeights, sumOfWeights;
double[] weightOfInstancesInClass, weightFactor, weightOfInstances;
Instances newData;
if (data.classIndex() < 0)
{
throw new System.Exception("Class index is not set!");
}
if (size() != data.numClasses())
{
throw new System.Exception("Misclassification cost matrix has " + "wrong format!");
}
weightFactor = new double[data.numClasses()];
weightOfInstancesInClass = new double[data.numClasses()];
for (int j = 0; j < data.numInstances(); j++)
{
//UPGRADE_WARNING: Data types in Visual C# might be different. Verify the accuracy of narrowing conversions. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1042'"
weightOfInstancesInClass[(int) data.instance(j).classValue()] += data.instance(j).weight();
}
sumOfWeights = Utils.sum(weightOfInstancesInClass);
// normalize the matrix if not already
for (int i = 0; i < size(); i++)
if (!Utils.eq(getXmlElement(i, i), 0))
{
CostMatrix normMatrix = new CostMatrix(this);
normMatrix.normalize();
return normMatrix.applyCostMatrix(data, random);
}
for (int i = 0; i < data.numClasses(); i++)
{
// Using Kai Ming Ting's formula for deriving weights for
// the classes and Breiman's heuristic for multiclass
// problems.
sumOfMissClassWeights = 0;
for (int j = 0; j < data.numClasses(); j++)
{
if (Utils.sm(getXmlElement(i, j), 0))
{
throw new System.Exception("Neg. weights in misclassification " + "cost matrix!");
}
sumOfMissClassWeights += getXmlElement(i, j);
}
weightFactor[i] = sumOfMissClassWeights * sumOfWeights;
sumOfWeightFactors += sumOfMissClassWeights * weightOfInstancesInClass[i];
}
for (int i = 0; i < data.numClasses(); i++)
{
weightFactor[i] /= sumOfWeightFactors;
}
// Store new weights
weightOfInstances = new double[data.numInstances()];
for (int i = 0; i < data.numInstances(); i++)
{
//UPGRADE_WARNING: Data types in Visual C# might be different. Verify the accuracy of narrowing conversions. "ms-help://MS.VSCC.v80/dv_commoner/local/redirect.htm?index='!DefaultContextWindowIndex'&keyword='jlca1042'"
weightOfInstances[i] = data.instance(i).weight() * weightFactor[(int) data.instance(i).classValue()];
}
// Change instances weight or do resampling
if (random != null)
{
return data.resampleWithWeights(random, weightOfInstances);
}
else
{
Instances instances = new Instances(data);
for (int i = 0; i < data.numInstances(); i++)
{
instances.instance(i).Weight = weightOfInstances[i];
}
return instances;
}
}
/// <summary> Filters an entire set of instances through a filter and returns
/// the new set.
///
/// </summary>
/// <param name="data">the data to be filtered
/// </param>
/// <param name="filter">the filter to be used
/// </param>
/// <returns> the filtered set of data
/// </returns>
/// <exception cref="Exception">if the filter can't be used successfully
/// </exception>
public static Instances useFilter(Instances data, Filter filter)
{
/*
System.err.println(filter.getClass().getName()
+ " in:" + data.numInstances());
*/
for (int i = 0; i < data.numInstances(); i++)
{
filter.input(data.instance(i));
}
filter.batchFinished();
Instances newData = filter.getOutputFormat();
Instance processed;
while ((processed = filter.output()) != null)
{
newData.add(processed);
}
/*
System.err.println(filter.getClass().getName()
+ " out:" + newData.numInstances());
*/
return newData;
}
private void result_Click(object sender, EventArgs e)
{
ArrayList algorithms = new ArrayList();
algorithms.Add("Naive Bayes");
algorithms.Add("K Nearest Neighbor");
algorithms.Add("Decision Tree");
algorithms.Add("Neural Network");
algorithms.Add("Support Vector Machine");
ArrayList successPercent = new ArrayList();
double res_Naive, res_KNN, res_NN, res_Tree, res_SVM = 0.0;
string nameOfAlgo = "";
//NAIVE BAYES ALGORITHM
weka.core.Instances insts = new weka.core.Instances(new java.io.FileReader(fileDirectory));
//CREATIING DYNAMIC GRIDVIEW FOR ADDING NEW INSTANCE
dataGridView1.ColumnCount = 2;
dataGridView1.RowCount = insts.numAttributes();
String[,] matrixOfInstances = new String[insts.numInstances(), insts.numAttributes()];
for (int y = 0; y < insts.numAttributes() - 1; y++)
{
dataGridView1.Rows[y].Cells[0].Value = insts.attribute(y).name();
if (insts.attribute(y).isNominal())
{
//nominalDataValues.Add(insts.attribute(y).toString());
string phrase = insts.attribute(y).toString();
string[] first = phrase.Split('{');
string[] second = first[1].Split('}');
string[] attributeValues = second[0].Split(',');
DataGridViewComboBoxCell comboColumn = new DataGridViewComboBoxCell();
foreach (var a in attributeValues)
{
comboColumn.Items.Add(a);
}
dataGridView1.Rows[y].Cells[1] = comboColumn;
}
}
insts.setClassIndex(insts.numAttributes() - 1);
cl_Naive = new weka.classifiers.bayes.NaiveBayes();
weka.filters.Filter myNominalData = new weka.filters.unsupervised.attribute.Discretize();
myNominalData.setInputFormat(insts);
insts = weka.filters.Filter.useFilter(insts, myNominalData);
//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_Naive.buildClassifier(train);
string str = cl_Naive.toString();
int numCorrect = 0;
for (int i = trainSize; i < insts.numInstances(); i++)
{
weka.core.Instance currentInst = insts.instance(i);
double predictedClass = cl_Naive.classifyInstance(currentInst);
if (predictedClass == insts.instance(i).classValue())
{
numCorrect++;
}
}
res_Naive = (double)((double)numCorrect / (double)testSize * 100.0);
successPercent.Add(res_Naive);
//kNN
weka.core.Instances insts2 = new weka.core.Instances(new java.io.FileReader(fileDirectory));
insts2.setClassIndex(insts2.numAttributes() - 1);
cl_Knn = new weka.classifiers.lazy.IBk();
//Nominal to Binary
weka.filters.Filter myBinaryData = new weka.filters.unsupervised.attribute.NominalToBinary();
myBinaryData.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myBinaryData);
//Normalization
weka.filters.Filter myNormalized = new weka.filters.unsupervised.instance.Normalize();
myNormalized.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myNormalized);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom2 = new weka.filters.unsupervised.instance.Randomize();
myRandom2.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myRandom2);
int trainSize2 = insts2.numInstances() * percentSplit / 100;
int testSize2 = insts2.numInstances() - trainSize2;
weka.core.Instances train2 = new weka.core.Instances(insts2, 0, trainSize2);
cl_Knn.buildClassifier(train2);
string str2 = cl_Knn.toString();
int numCorrect2 = 0;
for (int i = trainSize2; i < insts2.numInstances(); i++)
{
weka.core.Instance currentInst2 = insts2.instance(i);
double predictedClass = cl_Knn.classifyInstance(currentInst2);
if (predictedClass == insts2.instance(i).classValue())
{
numCorrect2++;
}
}
res_KNN = (double)((double)numCorrect2 / (double)testSize2 * 100.0);
successPercent.Add(res_KNN);
//Decision tree
weka.core.Instances insts3 = new weka.core.Instances(new java.io.FileReader(fileDirectory));
insts3.setClassIndex(insts3.numAttributes() - 1);
cl_Tree = new weka.classifiers.trees.J48();
weka.filters.Filter myNormalized2 = new weka.filters.unsupervised.instance.Normalize();
myNormalized2.setInputFormat(insts3);
insts3 = weka.filters.Filter.useFilter(insts3, myNormalized2);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom3 = new weka.filters.unsupervised.instance.Randomize();
myRandom3.setInputFormat(insts3);
insts3 = weka.filters.Filter.useFilter(insts3, myRandom3);
int trainSize3 = insts3.numInstances() * percentSplit / 100;
int testSize3 = insts3.numInstances() - trainSize3;
weka.core.Instances train3 = new weka.core.Instances(insts3, 0, trainSize3);
cl_Tree.buildClassifier(train3);
string str3 = cl_Tree.toString();
int numCorrect3 = 0;
for (int i = trainSize3; i < insts3.numInstances(); i++)
{
weka.core.Instance currentInst3 = insts3.instance(i);
double predictedClass = cl_Tree.classifyInstance(currentInst3);
if (predictedClass == insts3.instance(i).classValue())
{
numCorrect3++;
}
}
res_Tree = (double)((double)numCorrect3 / (double)testSize3 * 100.0);
successPercent.Add(res_Tree);
//Neural Network
weka.core.Instances insts4 = new weka.core.Instances(new java.io.FileReader(fileDirectory));
insts4.setClassIndex(insts4.numAttributes() - 1);
cl_NN = new weka.classifiers.functions.MultilayerPerceptron();
//Nominal to Binary
weka.filters.Filter myBinaryData2 = new weka.filters.unsupervised.attribute.NominalToBinary();
myBinaryData2.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, myBinaryData2);
//Normalization
weka.filters.Filter myNormalized3 = new weka.filters.unsupervised.instance.Normalize();
myNormalized3.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, myNormalized3);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom4 = new weka.filters.unsupervised.instance.Randomize();
myRandom4.setInputFormat(insts4);
insts4 = weka.filters.Filter.useFilter(insts4, myRandom4);
int trainSize4 = insts4.numInstances() * percentSplit / 100;
int testSize4 = insts4.numInstances() - trainSize4;
weka.core.Instances train4 = new weka.core.Instances(insts4, 0, trainSize4);
cl_NN.buildClassifier(train4);
string str4 = cl_NN.toString();
int numCorrect4 = 0;
for (int i = trainSize4; i < insts4.numInstances(); i++)
{
weka.core.Instance currentInst4 = insts4.instance(i);
double predictedClass = cl_NN.classifyInstance(currentInst4);
if (predictedClass == insts4.instance(i).classValue())
{
numCorrect4++;
}
}
res_NN = (double)((double)numCorrect4 / (double)testSize4 * 100.0);
successPercent.Add(res_NN);
//SVM
weka.core.Instances insts5 = new weka.core.Instances(new java.io.FileReader(fileDirectory));
insts5.setClassIndex(insts5.numAttributes() - 1);
cl_SVM = new weka.classifiers.functions.SMO();
//Nominal to Binary
weka.filters.Filter myBinaryData3 = new weka.filters.unsupervised.attribute.NominalToBinary();
myBinaryData3.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, myBinaryData3);
//Normalization
weka.filters.Filter myNormalized4 = new weka.filters.unsupervised.instance.Normalize();
myNormalized4.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, myNormalized4);
//randomize the order of the instances in the dataset.
weka.filters.Filter myRandom5 = new weka.filters.unsupervised.instance.Randomize();
myRandom5.setInputFormat(insts5);
insts5 = weka.filters.Filter.useFilter(insts5, myRandom5);
int trainSize5 = insts5.numInstances() * percentSplit / 100;
int testSize5 = insts5.numInstances() - trainSize5;
weka.core.Instances train5 = new weka.core.Instances(insts5, 0, trainSize5);
cl_SVM.buildClassifier(train5);
string str5 = cl_SVM.toString();
int numCorrect5 = 0;
for (int i = trainSize5; i < insts5.numInstances(); i++)
{
weka.core.Instance currentInst5 = insts5.instance(i);
double predictedClass = cl_SVM.classifyInstance(currentInst5);
if (predictedClass == insts5.instance(i).classValue())
{
numCorrect5++;
}
}
res_SVM = (double)((double)numCorrect5 / (double)testSize5 * 100.0);
successPercent.Add(res_SVM);
for (int i = 0; i < successPercent.Count; i++)
{
if ((double)successPercent[i] > max)
{
max = (double)successPercent[i];
count = i + 1;
}
}
for (int i = 0; i < count; i++)
{
nameOfAlgo = (string)algorithms[i];
}
textBox1.Text = nameOfAlgo + " is the most successful algorithm for this data set." + "(" + max + "%)\n";
}
public override double classifyInstance(weka.core.Instance instance)
{
if (m_instances.numInstances() == 0)
{
return(2);
}
if (m_instances.numAttributes() != instance.numAttributes())
{
throw new AssertException("different attribute.");
}
int n = (instance.numAttributes() - 1) / 2;
List <Tuple <int, int> > dist = new List <Tuple <int, int> >();
for (int i = 0; i < m_instances.numInstances(); ++i)
{
int d1 = 0, d2 = 0;
weka.core.Instance instanceI = m_instances.instance(i);
for (int j = 0; j < n; ++j)
{
//d += (int)((instanceI.value(j) - instance.value(j)) * (instanceI.value(j) - instance.value(j)));
if (instanceI.value(j) != instance.value(j))
{
if (instance.value(j) == 2 || instanceI.value(j) == 2)
{
d1++;
}
else
{
d1 += 4;
}
}
}
for (int j = n; j < 2 * n; ++j)
{
//d += (int)((instanceI.value(j) - instance.value(j)) * (instanceI.value(j) - instance.value(j)));
if (instanceI.value(j) != instance.value(j))
{
if (instance.value(j) == 2 || instanceI.value(j) == 2)
{
d2++;
}
else
{
d2 += 4;
}
}
}
int c = (int)instanceI.classValue();
//if (c == 0)
//{
// if (d1 < n / 4 && d1 < d2)
// {
// dist.Add(new Tuple<int, int>(d1, c));
// }
//}
//else if (c == 1)
//{
// if (d2 < n / 4 && d2 < d1)
// {
// dist.Add(new Tuple<int, int>(d2, c));
// }
//}
//else
//{
// throw new AssertException("");
//}
dist.Add(new Tuple <int, int>(d1 + d2, c));
}
if (dist.Count == 0)
{
return(2);
}
dist.Sort(new Comparison <Tuple <int, int> >((x, y) =>
{
return(x.Item1.CompareTo(y.Item1));
}));
int sum = 0, count = 0;
for (int i = 0; i < dist.Count; ++i)
{
if (dist[i].Item1 < n / 4 * 2 * 4)
{
if (dist[i].Item2 != 2 && dist[i].Item2 != 3)
{
sum += dist[i].Item2;
count++;
}
else
{
}
}
else
{
break;
}
}
if (count == 0)
{
return(2);
}
if (count < m_instances.numInstances() / 30)
{
return(2);
}
return((int)Math.Round((double)sum / count));
}
