//Přesunout MeansPointy do správných clusterů
private bool UpdateClusterMembership()
{
MinIndex min = new MinIndex();
bool changed = false;
double[] distances = new double[_numberOfClusters];
double[] sse_count = new double[_numberOfClusters];
double result = 0;
for (int i = 0; i < _normalizedDataToCluster.Count; ++i)
{
for (int k = 0; k < _numberOfClusters; ++k)
{
distances[k] = ElucDist.Get(_normalizedDataToCluster[i], _clusters[k]);
}
int newClusterId = min.Get(distances);
if (newClusterId != _normalizedDataToCluster[i].Cluster)
{
changed = true;
_normalizedDataToCluster[i].Cluster = _rawDataToCluster[i].Cluster = newClusterId;
if (print)
{
Console.WriteLine("Width: " + _rawDataToCluster[i].Width + ", Lenght: " +
_rawDataToCluster[i].Length + " ----> Cluster " + newClusterId);
}
}
//SSE pro cluster
sse_count[newClusterId] = sse_count[newClusterId] + distances.Min();
}
//SSE celkové
for (int o = 0; o < sse_count.Length; o++)
{
result = result + sse_count[o];
}
sse = result;
if (changed == false)
{
return(false);
}
if (IsClusterEmpty(_normalizedDataToCluster))
{
return(false);
}
return(true);
}
///<summary>
///Generate elucidan distance for the whole list of points.
///</summary>
public double[,] GenerateElucid(List <Iris> source)
{
ElucidanDistance elucid = new ElucidanDistance();
double[,] x = new double[source.Count, source.Count];
for (int i = 0; i < source.Count(); i++)
{
for (int z = 0; z < source.Count(); z++)
{
x[i, z] = elucid.Get(source[i], source[z]);
}
}
return(x);
}
public void Classify(int neighborsNumber)
{
Console.WriteLine();
ElucidanDistance ElucDist = new ElucidanDistance();
this.K = neighborsNumber;
double correct = 0, testN = 0;
//Vytvořím si 2D pole pro uložení vzdáleností
double[][] distances = new double[trainingSetValues.Count][];
for (int i = 0; i < trainingSetValues.Count; i++)
{
distances[i] = new double[2];
}
// pro všechny prvky v tetsovací sadě
for (var test = 0; test < this.testSetValues.Count; test++)
{
//paralelně počitám euclid z testovacího prvku na všechny prvky v training sadě
Parallel.For(0, trainingSetValues.Count, index =>
{
var dist = ElucDist.Get(this.testSetValues[test], this.trainingSetValues[index]);
distances[index][0] = dist;
distances[index][1] = index;
}
);
Console.WriteLine("--VALUE {0} closest {1} neighbors--", test, this.K);
// seřadit vzdalenosti ke všem sousedum v training sadě a vzít první K
var sortedDistances = distances.AsParallel().OrderBy(t => t[0]).Take(this.K);
string realClass = testSetClasses[test];
foreach (var d in sortedDistances)
{
string predictedClass = trainingSetClasses[(int)d[1]];
Console.Write("PREDICTED: {0}", predictedClass);
if (string.Equals(realClass, predictedClass) == true)
{
Console.Write(" - TRUE");
correct++;
}
testN++;
if (string.Equals(realClass, predictedClass) != true)
{
Console.Write(" - WRONG!");
}
Console.WriteLine();
}
Console.WriteLine();
}
Console.WriteLine("TOTAL: {0}", testN);
Console.WriteLine("CORRECT: {0}", correct);
Console.WriteLine("{0}%", (correct / testN) * 100);
Console.ReadLine();
Console.Clear();
Console.WriteLine();
}