public int ClassifyInstance(double[] attributes, out double[] percentages)
{
double classificationResult = 1.0;
testInstance.setDataset(dataSet);
testInstance.setClassMissing();
dataSet.add(testInstance);
for (int i = 0; i < attributes.Length; i++)
{
testInstance.setValue(i, attributes[i]);
}
classificationResult = m_cl.classifyInstance(testInstance);
dataSet.delete(0);
percentages = m_cl.distributionForInstance(testInstance);
return((int)classificationResult);
}
/// <summary>
/// Uses the classifier to classify an instance (from its featureValues).
/// </summary>
/// <param name="featureValues">An array of doubles that describe the instance.</param>
/// <returns>The string name of the classification of the instance.</returns>
public string classify(double[] featureValues)
{
//if (!classifierBuilt) { _classifier.buildClassifier(_dataSet); classifierBuilt = true; }
weka.core.Instance inst = new weka.core.Instance(1, featureValues);
inst.setDataset(_dataSet);
double result = _classifier.classifyInstance(inst);
weka.core.Attribute attribute = _dataSet.attribute(_dataSet.numAttributes() - 1);
string resultName = attribute.value((int)result);
// Get rid of this line once ARFF files are rewritten
if (resultName == "Label")
{
resultName = "Text";
}
//Console.WriteLine(resultName);
return(resultName);
}
private void testButton_Click(object sender, EventArgs e)
{
var form = Form.ActiveForm as Form1;
if (readyToTest)
{
weka.classifiers.Classifier cl = classifiers[highestSuccessRate.Key];
weka.core.Instance inst = new weka.core.Instance(insts.numAttributes() - 1);
inst.setDataset(insts);
for (int i = 0; i < inputObjects.Count; i++)
{
if (inputObjects[i].numeric)
{
inst.setValue(i, Decimal.ToDouble(inputObjects[i].num.Value));
}
else
{
inst.setValue(i, inputObjects[i].nom.SelectedItem.ToString());
}
}
try
{
string[] values = insts.attribute(insts.numAttributes() - 1).toString().Split('{', '}')[1].Split(',');
double classOfData = cl.classifyInstance(inst);
int idx = Convert.ToInt32(classOfData);
form.testResult.Text = values[idx];
}
catch (Exception ex)
{
MessageBox.Show(ex.Message, "Error", MessageBoxButtons.OK, MessageBoxIcon.Error);
}
}
else
{
MessageBox.Show("Program is not ready to test, probably needs to process data first.", "Not ready", MessageBoxButtons.OK, MessageBoxIcon.Information);
}
}
public double[] ClassifyNewData(double[] newData)
{
weka.core.Instance newInstance = new weka.core.Instance(1, newData);
newInstance.setDataset(playerData);
return(classifier.distributionForInstance(newInstance));
}
public weka.core.Instance GenerateFeatures(PokerHand hand, int rIdx, int aIdx, weka.core.Instances data, bool generateClass = true)
{
// Check that we are using limit betting.
Debug.Assert(hand.Context.BettingType == BettingType.FixedLimit);
var results = new weka.core.Instance(data.numAttributes());
results.setDataset(data);
int attIdx = 0;
foreach (var method in typeof(LimitFeatureGenerator).GetMethods())
{
// Get all the features of this class.
var attributes = method.GetCustomAttributes(typeof(Feature), true);
if (attributes.Length == 0)
continue;
// Get the feature attribute on this method.
var attr = ((Feature)attributes[0]);
// Get the name for this attribute
string name = attr.Name;
//Console.WriteLine("Hand: {0} Feature: {1}", hand.Context.ID, name);
// Get the feature only if it's applicable to this situation.
object feature = null;
if(rIdx >= (int)attr.MinRound && rIdx <= (int)attr.MaxRound)
feature = method.Invoke(this, new object[] { hand, rIdx, aIdx });
if (SkipMissingFeatures && (feature == null || feature.ToString() == "?"))
continue;
switch (attr.FType) {
case FeatureType.Continuous:
results.setValue(attIdx, (double)feature);
break;
case FeatureType.Discrete:
results.setValue(attIdx, (int)feature);
break;
case FeatureType.Boolean:
case FeatureType.Nominal:
case FeatureType.Enum:
{
var attribute = data.attribute(attIdx);
var attVal = attribute.indexOfValue(feature.ToString());
if(attVal < 0 || attVal > attribute.numValues())
throw new Exception(string.Format ("Invalid attribute value: {0} for attribute {1} (idx: {2} total values: {3}", feature.ToString(), name, attVal, attribute.numValues()));
results.setValue(attribute, attVal);
}
break;
case FeatureType.String:
{
var attribute = data.attribute(attIdx);
results.setValue(attribute, feature.ToString());
}
break;
default: throw new Exception("Unspecified feature type for feature: " + method.Name);
}
attIdx++;
}
if(generateClass)
{
var classAttr = data.classAttribute();
switch (hand.Rounds[rIdx].Actions[aIdx].Type)
{
case ActionType.Bet:
case ActionType.Raise: results.setClassValue(classAttr.indexOfValue("Raise"));
break;
case ActionType.Call:
case ActionType.Check: results.setClassValue(classAttr.indexOfValue("Call"));
break;
case ActionType.Fold: results.setClassValue(classAttr.indexOfValue("Fold"));;
break;
default:
break;
}
}
return results;
}
public override double classifyInstance(Instance instance)
{
if (m_mustValue.HasValue)
return m_mustValue.Value;
instance.setDataset(m_sampleInstances);
instance.setClassValue(0);
m_sampleInstances.clear();
m_sampleInstances.add(instance);
double[] d = classifyInstances(m_sampleInstances);
return d[0];
}
private void btnDiscover_Click(object sender, EventArgs e)
{
string type = model.GetType().ToString();
bool flag = false;
bool flag2 = false;
//input kontrolleri
if (nominal != null)
{
for (int i = 0; i < nominal.Length; i++)
{
if (nominal[i].SelectedIndex == -1)
{
flag = true;
break;
}
}
}
if (numeric != null)
{
for (int i = 0; i < numeric.Length; i++)
{
if (String.IsNullOrEmpty(numeric[i].Text))
{
flag2 = true;
break;
}
}
}
if (numAtt == numeric.Length && flag2 == true)
{
MessageBox.Show("Please select value!", "Error Message!");
}
else if (numAtt == nominal.Length && flag == true)
{
MessageBox.Show("Please select value!", "Error Message!");
}
else if ((nominal.Length + numeric.Length) == numAtt && (flag == true || flag2 == true))
{
MessageBox.Show("Please select value!", "Error Message!");
}
else
{
weka.core.Instance newIns = new weka.core.Instance(numAtt + 1);
newIns.setDataset(insts);
int i1 = 0, i2 = 0;
for (int i = 0; i < numAtt; i++)
{
//nominal
if (typeAtt[i])
{
newIns.setValue(i, nominal[i1].SelectedItem.ToString());
i1++;
}
//numeric
else
{
newIns.setValue(i, double.Parse(numeric[i2].Text));
i2++;
}
}
weka.core.Instances insts2 = new weka.core.Instances(insts);
insts2.add(newIns);
if (type == "weka.classifiers.bayes.NaiveBayes")
{
weka.filters.Filter myDiscretize = new weka.filters.unsupervised.attribute.Discretize();
myDiscretize.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myDiscretize);
}
else if (type == "weka.classifiers.functions.Logistic")
{
weka.filters.Filter myDummy = new weka.filters.unsupervised.attribute.NominalToBinary();
myDummy.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myDummy);
weka.filters.Filter myNormalize = new weka.filters.unsupervised.instance.Normalize();
myNormalize.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myNormalize);
}
else if (type == "new weka.classifiers.lazy.IBk")
{
weka.filters.Filter myDummy = new weka.filters.unsupervised.attribute.NominalToBinary();
myDummy.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myDummy);
weka.filters.Filter myNormalize = new weka.filters.unsupervised.instance.Normalize();
myNormalize.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myNormalize);
}
else if (type == "weka.classifiers.trees.J48")
{
}
else if (type == "weka.classifiers.trees.RandomForest")
{
}
else if (type == "weka.classifiers.trees.RandomTree")
{
}
else if (type == "weka.classifiers.functions.MultilayerPerceptron")
{
weka.filters.Filter myDummy = new weka.filters.unsupervised.attribute.NominalToBinary();
myDummy.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myDummy);
weka.filters.Filter myNormalize = new weka.filters.unsupervised.instance.Normalize();
myNormalize.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myNormalize);
}
else if (type == "weka.classifiers.functions.SMO")
{
weka.filters.Filter myDummy = new weka.filters.unsupervised.attribute.NominalToBinary();
myDummy.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myDummy);
weka.filters.Filter myNormalize = new weka.filters.unsupervised.instance.Normalize();
myNormalize.setInputFormat(insts2);
insts2 = weka.filters.Filter.useFilter(insts2, myNormalize);
}
double index = model.classifyInstance(insts2.lastInstance());
//Model okuma kısmı
weka.classifiers.Classifier cls = (weka.classifiers.Classifier)weka.core.SerializationHelper.read("models/mdl.model");
lblResult2.Text = "Result= " + insts2.attribute(insts2.numAttributes() - 1).value(Convert.ToInt16(index));
}
}
public double[] ClassifyNewData(double[] newData)
{
weka.core.Instance newInstance = new weka.core.Instance(1,newData);
newInstance.setDataset(playerData);
return classifier.distributionForInstance(newInstance);
}
