public static List <TestCentroid> GetClustersTest(List <TestCentroid> clusters, float alpha, List <DocumentVectorTest> data)
{
List <TestCentroid> newclusters = new List <TestCentroid>();
int N = data.Count;
int number_of_clusters = clusters.Count;
float MIN_POINTS = alpha * N;
for (int i = 0; i < number_of_clusters; i++)
{
TestCentroid centroid = new TestCentroid
{
GroupedDocument = new List <DocumentVectorTest>()
};
if (clusters[i].GroupedDocument.Count >= MIN_POINTS)
{
foreach (var elements in clusters[i].GroupedDocument)
{
centroid.GroupedDocument.Add(elements);
newclusters.Add(centroid);
}
}
}
return(newclusters);
}
public static Tuple <int[], int[], List <TestCentroid> > Set(List <DocumentVectorTest> docCollection)
{
Tuple <int[], int[], List <TestCentroid> > result;
parent = new int[docCollection.Count];
rank = new int[docCollection.Count];
var cntroidSet = new List <TestCentroid>();
for (int i = 0; i < docCollection.Count; i++)
{
parent[i] = i;
rank[i] = 0;
}
TestCentroid newCentroid;
//here is a problem cntroidSet.Count must be 46 not 23!!!
List <DocumentVectorTest> docCollectionCopy = new List <DocumentVectorTest>(docCollection);
for (int j = 0; j < docCollectionCopy.Count; j++)
{
newCentroid = new TestCentroid();
newCentroid.GroupedDocument = new List <DocumentVectorTest>();
newCentroid.GroupedDocument.Add(docCollectionCopy[j]);
cntroidSet.Add(newCentroid);
}
result = new Tuple <int[], int[], List <TestCentroid> >(parent, rank, cntroidSet);
return(result);
}
public static List <TestCentroid> CentroidCalculationsForKMeans(List <DocumentVectorTest> data, int ClusterNumber)
{
List <TestCentroid> centroidList = new List <TestCentroid>();
Random randomizer = new Random();
HashSet <int> indexSet = new HashSet <int>();
int index = 0;
while (centroidList.Count != ClusterNumber)
{
index = randomizer.Next(0, data.Count + 1);
if (!indexSet.Contains(index))
{
indexSet.Add(index);
TestCentroid newCentroid = new TestCentroid();
newCentroid.GroupedDocument = new List <DocumentVectorTest>();
newCentroid.GroupedDocument.Add(data[index]);
centroidList.Add(newCentroid);
}
else if (indexSet.Contains(index))
{
continue;
}
}
foreach (var doc in centroidList)
{
doc.CalculateMeans();
doc.GroupedDocument.Clear();
}
return(centroidList);
}
public static float[] CalculateProbabilityArray_Test(TestCentroid oldCentroid, List <DocumentVectorTest> vSpace)
{
List <DocumentVectorTest> vSpaceCopy = new List <DocumentVectorTest>(vSpace);
float[] vector_A = oldCentroid.GroupedDocument[0].VectorSpace;
float[] DistanceQuad = new float[vSpaceCopy.Count];
for (int i = 0; i < DistanceQuad.Length; i++)
{
DistanceQuad[i] = 0;
}
float SumDistanceQuad = 0;
int previous_index = vSpaceCopy.IndexOf(oldCentroid.GroupedDocument[0]);
for (int j = 0; j <= vSpaceCopy.Count - 1; j++)
{
float[] vector_B = vSpaceCopy[j].VectorSpace;
for (int k = 0; k <= vSpaceCopy[j].VectorSpace.Length - 1; k++)
{
DistanceQuad[j] += (float)Math.Pow((vector_A[k] - vector_B[k]), 2);
}
SumDistanceQuad += DistanceQuad[j];
}
for (int j = 0; j <= DistanceQuad.Length - 1; j++)
{
DistanceQuad[j] = DistanceQuad[j] / SumDistanceQuad;
}
return(DistanceQuad);
}
public static List <TestCentroid> CreateClusterSet(int clusterNumber)
{
List <TestCentroid> result = new List <TestCentroid>();
for (int i = 0; i < clusterNumber; i++)
{
TestCentroid centroid = new TestCentroid();
centroid.GroupedDocument = new List <DocumentVectorTest>();
result.Add(centroid);
}
return(result);
}
private static List <TestCentroid> TestCentroidInitializer(int numClusters)
{
List <TestCentroid> initializedList = new List <TestCentroid>(numClusters);
for (int i = 0; i < numClusters; i++)
{
TestCentroid newTestCentroid = new TestCentroid();
newTestCentroid.GroupedDocument = new List <DocumentVectorTest>();
initializedList.Add(newTestCentroid);
}
return(initializedList);
}
private static TestCentroid Calculate_Next_Centroid_Test(TestCentroid firstcentroid, List <DocumentVectorTest> vSpace)
{
TestCentroid next_centroid = new TestCentroid();
next_centroid.GroupedDocument = new List <DocumentVectorTest>();
List <DocumentVectorTest> vSpaceCopy = new List <DocumentVectorTest>(vSpace);
float[] probabilitiesMatrixSimple = CalculateProbabilityArray_Test(firstcentroid, vSpaceCopy);
float[] probabilitiesMatrix = new float[probabilitiesMatrixSimple.Length];
for (var i = 0; i < probabilitiesMatrix.Length; i++)
{
probabilitiesMatrix[i] = 0;
}
for (var i = 0; i < probabilitiesMatrix.Length; i++)
{
for (var j = 0; j < i; j++)
{
probabilitiesMatrix[i] += probabilitiesMatrixSimple[j];
}
}
Random rand = new Random();
float interval_Value = (float)rand.NextDouble();
float sum_Of_Probabilies = 0.0F;
int index_of_min_distance_element = 0;
for (int i = 0; i < probabilitiesMatrix.Length; i++)
{
sum_Of_Probabilies += probabilitiesMatrix[i];
//here are the problem! - trying to fix;
}
for (int j = 0; j < probabilitiesMatrix.Length; j++)
{
if (sum_Of_Probabilies > interval_Value & sum_Of_Probabilies < probabilitiesMatrix[j])
{
index_of_min_distance_element = j - 1;
}
else
{
continue;
}
}
next_centroid.GroupedDocument.Add(vSpaceCopy[index_of_min_distance_element]);
vSpaceCopy.RemoveAt(index_of_min_distance_element);
// but here we can find distance from oldCentroid tp old Centroid
return(next_centroid);
}
public static List <TestCentroid> CentroidCalculationsForTestKMeansPP(List <DocumentVectorTest> dataPP, int ClusterNumberPP)
{
List <TestCentroid> centroidListPP = new List <TestCentroid>();
List <DocumentVectorTest> dataPPCopy = new List <DocumentVectorTest>(dataPP);
List <DocumentVectorTest> existingCentroids = new List <DocumentVectorTest>();
Random randomizerPP = new Random();
float[] distances = new float[dataPP.Count];
int indexOfFirstElement = randomizerPP.Next(0, dataPP.Count);// + 1);
TestCentroid firstCentroid = new TestCentroid();
firstCentroid.GroupedDocument = new List <DocumentVectorTest>();
firstCentroid.GroupedDocument.Add(dataPP[indexOfFirstElement]);
centroidListPP.Add(firstCentroid);
HashSet <TestCentroid> stringHashSet = new HashSet <TestCentroid>();
while (centroidListPP.Count != ClusterNumberPP)
{
TestCentroid newCentroid = new TestCentroid();
newCentroid.GroupedDocument = new List <DocumentVectorTest>();
newCentroid = Calculate_Next_Centroid_Test(firstCentroid, dataPPCopy);
if (!existingCentroids.Contains(newCentroid.GroupedDocument[0]))
{
existingCentroids.Add(newCentroid.GroupedDocument[0]);
centroidListPP.Add(newCentroid);
//zmiana1
stringHashSet.Add(newCentroid);
firstCentroid = newCentroid;
dataPPCopy.Remove(newCentroid.GroupedDocument[0]);
}
//zmiana2
else if (existingCentroids.Contains(newCentroid.GroupedDocument[0]) || stringHashSet.Contains(newCentroid))
{
continue;
}
//zmiana 3
centroidListPP = stringHashSet.ToList();
}
return(centroidListPP);
}
public static List <TestCentroid> GetClustersTest(int[] clusters, float alpha, List <DocumentVectorTest> data)
{
List <TestCentroid> centroidSet = new List <TestCentroid>();
HashSet <int> clustersSet = new HashSet <int>();
int N = data.Count;
int number_of_clusters = clusters.Length;
float MIN_POINTS = alpha * N;
for (int i = 0; i < N; i++)
{
clustersSet.Add(clusters[i]);
}
for (int i = 0; i < clustersSet.Count; i++)
{
TestCentroid centroid = new TestCentroid
{
GroupedDocument = new List <DocumentVectorTest>()
};
var docIndex = clustersSet.ElementAt(i);
centroid.GroupedDocument.Add(data[docIndex]);
centroidSet.Add(centroid);
}
for (int j = 0; j < clustersSet.Count; j++)
{
for (int i = 0; i < N; i++)
{
if (clustersSet.ElementAt(j) == clusters[i])
{
centroidSet[j].GroupedDocument.Add(data[i]);
}
}
}
return(centroidSet);
}
