Three concentric rings of 300, 400 and 500 points with radii 0.5, 1.2 and 2.0, plus 60 uniformly scattered noise points. A standard case where density-based methods succeed and centroid-based ones cannot.
Format
A numeric matrix with 1260 rows and columns x and y. No labels
are included; the generating code is in data-raw/rings.R.
Examples
res <- shoal_hdbscan(rings, min_cluster_size = 15L, min_samples = 5L)
res
#>
#> ── HDBSCAN Clustering
#> Metric: "euclidean"
#> Parameters: alpha = 1, min_samples = 5, min_cluster_size = 15, boruvka = TRUE
#> Clusters: 3, Noise points: 26
#> GLOSH outlier scores: median 0.079, max 0.965