public VIC(Dataset dataset, IEnumerable <ISupervisedClassifier> supervisedClassifiers, IUnsupervisedClassifier clusteringAlgorithm)
{
OriginalDataset = dataset;
ClusteredDataset = null;
SupervisedClassifiers = supervisedClassifiers;
ClusteringAlgorithm = clusteringAlgorithm;
}
public Dataset GetClusteredDataset(Dataset dataset, IUnsupervisedClassifier clusteringAlgorithm)
{
if (dataset == null)
{
throw new ArgumentNullException("An initial dataset is required");
}
Header header = new Header($"Clustered {dataset.Header.RelationName}", dataset.Header.Features);
IEnumerable <Instance> instances = dataset.Instances;
instances.ToList().ForEach(i => i[dataset.Header.ClassFeature] = null);
Dataset clusteredDataset = new Dataset(header, instances.ToArray());
Feature classFeature = clusteredDataset.Header.ClassFeature;
List <Feature> features = clusteredDataset.Header.Features.ToList();
features.Remove(clusteredDataset.Header.ClassFeature);
clusteringAlgorithm.Initialize(new Dictionary <string, object> {
{ "k", ((NominalFeature)classFeature.Type).Values.Count }
});
clusteringAlgorithm.Train(clusteredDataset.GetColumns(features), clusteredDataset.GetClasses());
int[] clusteringResults = clusteringAlgorithm.Classify(clusteredDataset.GetColumns(features));
for (int i = 0; i < clusteringResults.Length; i++)
{
clusteredDataset.Instances[i][classFeature] = clusteringResults[i];
}
return(clusteredDataset);
}
