Merge pull request #109 from asturm-fe/107-dbscan-fill-pointlist-center

107: Fill cluster.PointList and cluster.Center in DBScan()

clusters/dbscan/dbscan.go

@@ -37,7 +37,8 @@ func DBScan(points clusters.PointList, eps float64, minPoints int) (clusterArray
 		if len(neighborPts) < minPoints {
 			noise = append(noise, i)
 		} else {
-			cluster := clusters.Cluster{C: C, Points: []int{i}}
+			// init cluster with center point
+			cluster := clusters.Cluster{C: C, Points: []int{i}, PointList: []space.Point{points[i]}}
 			members[i] = true
 			C++
 			// expandCluster goes here inline
@@ -63,9 +64,11 @@ func DBScan(points clusters.PointList, eps float64, minPoints int) (clusterArray
 
 				if !members[k] {
 					cluster.Points = append(cluster.Points, k)
+					cluster.PointList = append(cluster.PointList, points[k])
 					members[k] = true
 				}
 			}
+			cluster.Recenter()
 			clusterArray = append(clusterArray, cluster)
 		}
 	}

clusters/dbscan/dbscan_test.go

@@ -73,6 +73,7 @@ func TestRangeQueryKDTree(t *testing.T) {
 }
 
 func TestDBScan(t *testing.T) {
+	setup()
 	clusterArray, noise := DBScan(s.points, 0.8, 10)
 
 	goodClusters := []clusters.Cluster{
@@ -157,4 +158,36 @@ func TestDBScan(t *testing.T) {
 			return
 		}
 	}
+
+	// Verify cluster.PointList contains the correct points
+	for _, cluster := range clusterArray {
+		if len(cluster.Points) != len(cluster.PointList) {
+			t.Errorf("Cluster #%d cluster.Points contains %d points, while cluster.Pointlist contains %d points. Must be the same!", cluster.C, len(cluster.Points), len(cluster.PointList))
+			return
+		}
+		// cluster.Points has been sorted, but cluster.PointList has not so we have to loop cluster.PointList for each cluster.Point to find the match
+		// multiple matches are allowed as a cluster can contain multiple points with the same coordinates
+		matchedPoints := make(map[int]bool, len(cluster.Points))
+		for _, pointIdx := range cluster.Points {
+			for _, point := range cluster.PointList {
+				if reflect.DeepEqual(s.points[pointIdx], point) {
+					matchedPoints[pointIdx] = true
+				}
+			}
+		}
+		// every index in cluster.Points must have a match in cluster.PointList
+		for _, pointIdx := range cluster.Points {
+			if !matchedPoints[pointIdx] {
+				t.Errorf("Cluster #%d pointIdx %d (%v) has no match in cluster.PointList", cluster.C, pointIdx, s.points[pointIdx])
+				return
+			}
+		}
+	}
+
+	// Verify cluster.Center is filled
+	for _, cluster := range clusterArray {
+		if cluster.Center == nil || len(cluster.Center) == 0 {
+			t.Errorf("cluster.Center should not be empty")
+		}
+	}
 }