/**
* calculate the max cluster size
*
* @return max cluster size
*/
private double calculateMaxClusterSize() {
double maxClusterSize = Double.MIN_VALUE;
for (int i = 0; i < this.clusters.size(); i++) {
Cluster c = this.clusters.get(i);
if (!c.getIsActive() || c.getPointsIDs().size() < 30) continue;
double clusterSize = calculateClusterSize(c);
if (clusterSize > maxClusterSize) {
maxClusterSize = clusterSize;
}
}
return maxClusterSize;
}
/**
* calculate the minimum distance between two clusters
*
* @param ci cluster ci
* @param cj cluster cj
* @return min distance between ci and cj
*/
private double calculateMinDistanceBetweenTwoClusters(Cluster ci, Cluster cj) {
double minDist = 0;
ArrayList<Integer> ciPoints = ci.getPointsIDs();
ArrayList<Integer> cjPoints = cj.getPointsIDs();
for (int i = 0; i < ciPoints.size(); i++) {
DatasetPattern ciPoint = this.dataset.get(ciPoints.get(i));
for (int j = 0; j < cjPoints.size(); j++) {
DatasetPattern cjPoint = this.dataset.get(cjPoints.get(j));
double distance = EuclideanDistance.calculateDistance(ciPoint, cjPoint);
minDist += distance;
}
}
return minDist / ((ciPoints.size() - 1) * (cjPoints.size() * 1));
}
public DunnIndex(
ArrayList<Cluster> clusters,
ArrayList<DenseRegion> denseRegions,
ArrayList<DatasetPattern> dataset) {
this.dataset = dataset;
this.maxClusterSize = 0;
this.clusters = clusters;
for (int i = 0; i < clusters.size(); i++) {
Cluster c = this.clusters.get(i);
if (!c.getIsActive()) continue;
ArrayList<DenseRegion> clusterDenseRegions = c.getRegions();
for (int j = 0; j < clusterDenseRegions.size(); j++) {
DenseRegion d = clusterDenseRegions.get(j);
c.addPointsList(d.getPoints());
}
}
this.maxClusterSize = calculateMaxClusterSize();
}
/**
* Calculate the cluster size or diameter
*
* @param c cluster
* @return cluster size or diameter
*/
private double calculateClusterSize(Cluster c) {
double size = 0;
ArrayList<Integer> clusterPoints = c.getPointsIDs();
for (int i = 0; i < clusterPoints.size(); i++) {
DatasetPattern point1 = this.dataset.get(clusterPoints.get(i));
for (int j = 0; j < clusterPoints.size(); j++) {
if (i == j) continue;
DatasetPattern point2 = this.dataset.get(clusterPoints.get(j));
size += EuclideanDistance.calculateDistance(point1, point2);
}
}
return size / (clusterPoints.size() * (clusterPoints.size() - 1));
}
/**
* calculate the dunn index between two clusters
*
* @param clusters clusters
* @param dataset dataset
* @return dunn index
*/
public double calculateDunnIndex() {
double minDI = Double.MAX_VALUE;
for (int i = 0; i < clusters.size(); i++) {
Cluster ci = clusters.get(i);
if (!ci.getIsActive()) continue;
if (ci.getPointsIDs().size() < 30) continue;
for (int j = 0; j < clusters.size(); j++) {
Cluster cj = clusters.get(j);
if (ci.getID() == cj.getID() || !cj.getIsActive()) continue;
if (cj.getPointsIDs().size() < 30) continue;
double minDistBetCiCj = calculateMinDistanceBetweenTwoClusters(ci, cj);
double DIij = minDistBetCiCj / this.maxClusterSize;
if (DIij < minDI) {
minDI = DIij;
}
}
}
return minDI;
}