private Classifier ClassifyCustom()
{
List <ObjectModel> objectsToRemove = new List <ObjectModel>();
var dbTrain = new Database();
var dbTest = new Database();
foreach (var obj in _database.Objects)
{
dbTrain.AddObject(obj);
}
dbTrain.FeaturesIDs = _database.FeaturesIDs;
dbTest.AddObject(new ObjectModel("A", new List <float> {
1, 1
}));
_classifier = new Classifier(dbTrain, dbTest);
_classifier.ClassifyNearestNeighbour();
_classifier.ClassifyKNearestNeighbour(uint.Parse(TextBoxNeighbours.Text));
_classifier.ClassifyNearesMean();
_classifier.ClassifyKNearestMean(int.Parse(TextBoxMeans.Text));
return(_classifier);
}
private Classifier ClassifyGiven()
{
List <ObjectModel> objectsToRemove = new List <ObjectModel>();
var dbTrain = new Database();;
var dbTest = new Database();
foreach (var obj in _database.Objects)
{
dbTrain.AddObject(obj);
}
int objectsQuant = dbTrain.Objects.Count * int.Parse(TextBoxTestObjects.Text) / 100;
dbTrain.FeaturesIDs = _database.FeaturesIDs;
for (int i = 0; i < objectsQuant; i++)
{
int r = _random.Next(objectsQuant);
dbTest.AddObject(_database.Objects[r]);
objectsToRemove.Add(_database.Objects[r]);
}
foreach (var obj in objectsToRemove)
{
dbTrain.RemoveObject(obj);
}
_classifier = new Classifier(dbTrain, dbTest);
_classifier.ClassifyNearestNeighbour();
_classifier.ClassifyKNearestNeighbour(uint.Parse(TextBoxNeighbours.Text));
_classifier.ClassifyNearesMean();
_classifier.ClassifyKNearestMean(int.Parse(TextBoxMeans.Text));
return(_classifier);
}
private void ButtonClassify2_Click(object sender, RoutedEventArgs e)
{
int k = int.Parse(TextBoxParts.Text);
List <ObjectModel> objectsToRemove = new List <ObjectModel>();
var dbTrain = new Database();
var dbTest = new Database();
foreach (var obj in _database.Objects)
{
dbTrain.AddObject(obj);
}
int objectsQuant = dbTrain.Objects.Count;
int partQuantity = objectsQuant / k;
IList <List <ObjectModel> > parts = new List <List <ObjectModel> >();
var classAobjects = dbTrain.Objects.Where(o => o.ClassName.Equals(dbTrain.ClassNames[0])).ToList();
var classBobjects = dbTrain.Objects.Where(o => o.ClassName.Equals(dbTrain.ClassNames[1])).ToList();
for (int i = 0; i < k; i++)
{
var part = new List <ObjectModel>();
var objAforPart = classAobjects.Skip(classAobjects.Count / k * i).Take(classAobjects.Count / k).ToList();
var objBforPart = classBobjects.Skip(classBobjects.Count / k * i).Take(classBobjects.Count / k).ToList();
foreach (var obj in objAforPart)
{
part.Add(obj);
}
foreach (var obj in objBforPart)
{
part.Add(obj);
}
parts.Add(part);
}
float tempResultNN = 0f, tempResultKNN = 0f, tempResultNM = 0f, tempResultKNM = 0f;
foreach (var part in parts)
{
dbTest = new Database();
dbTrain = new Database();
dbTrain.FeaturesIDs = _database.FeaturesIDs;
dbTrain.ClassNames = _database.ClassNames;
foreach (var obj in part)
{
dbTest.Objects.Add(obj);
}
var trainingParts = parts.Where(p => p != part).ToList();
foreach (var trainingPart in trainingParts)
{
foreach (var obj in trainingPart)
{
dbTrain.Objects.Add(obj);
}
}
_classifier = new Classifier(dbTrain, dbTest);
_classifier.ClassifyNearestNeighbour();
_classifier.ClassifyKNearestNeighbour(uint.Parse(TextBoxNeighbours.Text));
_classifier.ClassifyNearesMean();
_classifier.ClassifyKNearestMean(int.Parse(TextBoxMeans.Text));
tempResultNN += _classifier.NearestNeighbourResult.Effectiveness;
tempResultKNN += _classifier.KNearestNeighbourResult.Effectiveness;
tempResultNM += _classifier.NearestMeanResult.Effectiveness;
tempResultKNM += _classifier.KNearestMeanResult.Effectiveness;
}
float finalResultNN = tempResultNN / k;
float finalResultKNN = tempResultKNN / k;
float finalResultNM = tempResultNM / k;
float finalResultKNM = tempResultKNM / k;
List <MethodResult2> results = new List <MethodResult2>();
results.Add(new MethodResult2 {
MethodName = "NN", Effectiveness = finalResultNN
});
results.Add(new MethodResult2 {
MethodName = "KNN", Effectiveness = finalResultKNN
});
results.Add(new MethodResult2 {
MethodName = "NM", Effectiveness = finalResultNM
});
results.Add(new MethodResult2 {
MethodName = "KNM", Effectiveness = finalResultKNM
});
DataGridClassification2.ItemsSource = results;
}