public IClusteringCentroid <DomainType> Cluster(IDataSet <DomainType> data_set)
{
IFunction <IList <DomainType[]>, ICentroidDistance <DomainType> > centroid_calculator = centroid_calculator_template.Generate(data_set.DataContext);
IList <DomainType[]> instance_features_list = data_set.FeatureData;
RandomNumberGenerator generator = new RNGCryptoServiceProvider();
IList <ICentroidDistance <DomainType> > centroids = new List <ICentroidDistance <DomainType> >();
int[] permutation = generator.RandomPermutation(instance_features_list.Count);
for (int centroid_index = 0; centroid_index < this.desired_cluster_count; centroid_index++)
{
IList <DomainType[]> centroid_members = new List <DomainType[]>();
centroid_members.Add(instance_features_list[permutation[centroid_index]]);
if (centroid_members.Count != 0)
{
centroids.Add(centroid_calculator.Compute(centroid_members));
}
}
Cluster(instance_features_list, centroid_calculator, centroids);
return(new ClusteringCentroid <DomainType>(data_set.DataContext, centroids));
}
public IClusteringCentroid <DomainType> Cluster(IDataSet <DomainType> data_set)
{
IFunction <IList <DomainType[]>, ICentroidDistance <DomainType> > centroid_calculator = centroid_calculator_template.Generate(data_set.DataContext);
IList <DomainType[]> instance_features_list = data_set.FeatureData;
IList <IList <DomainType[]> > cluster_members = new List <IList <DomainType[]> >();
IList <ICentroidDistance <DomainType> > centroids = new List <ICentroidDistance <DomainType> >();
// add first cluster
cluster_members.Add(new List <DomainType[]>());
cluster_members[0].Add(instance_features_list[0]);
centroids.Add(centroid_calculator.Compute(cluster_members[0]));
// assign to clusters
for (int index_instance = 1; index_instance < instance_features_list.Count; index_instance++)
{
DomainType[] instance_features = instance_features_list[index_instance];
int best_cluster_index = 0;
double best_distance = centroids[0].ComputeDistance(instance_features);
for (int cluster_index = 1; cluster_index < centroids.Count; cluster_index++)
{
double distance = centroids[cluster_index].ComputeDistance(instance_features);
if (distance.CompareTo(best_distance) == 1)
{
best_distance = distance;
best_cluster_index = cluster_index;
}
}
if (best_distance.CompareTo(critical_distance) == 1)
{
// create new cluster
IList <DomainType[]> new_cluster = new List <DomainType[]>();
new_cluster.Add(instance_features);
cluster_members.Add(new_cluster);
centroids.Add(centroid_calculator.Compute(new_cluster));
}
else
{
// add to existing cluster and recompute centroid
cluster_members[best_cluster_index].Add(instance_features);
centroids[best_cluster_index] = centroid_calculator.Compute(cluster_members[best_cluster_index]);
}
}
return(new ClusteringCentroid <DomainType>(data_set.DataContext, centroids));
}
