HDBSCAN<T>

Models & Types · AiDotNet.Clustering.Density

HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise).

Inheritance: ClusteringBase<T> → HDBSCAN<T>

For Beginners

HDBSCAN automatically finds the right density level.

The key insight:

  • Instead of picking ONE epsilon (like DBSCAN)
  • Look at how clusters form at ALL epsilon values
  • Pick clusters that “live longest” in the hierarchy

Core distance: How far to reach k nearest neighbors Mutual reachability: Max of core distances and actual distance This makes sparse points “push away” from dense clusters.

Benefits:

  • No epsilon to tune
  • Finds varying-density clusters
  • Robust noise detection
  • Provides cluster hierarchy

How It Works

HDBSCAN extends DBSCAN by finding clusters at all density levels and selecting the most stable clusters. It uses mutual reachability distance and minimum spanning tree construction.

Algorithm steps:

  1. Compute core distances (distance to k-th nearest neighbor)
  2. Build mutual reachability graph
  3. Construct minimum spanning tree
  4. Build cluster hierarchy (condensed tree)
  5. Extract clusters using stability

Example

using AiDotNet;
using AiDotNet.Data.Loaders;
using AiDotNet.Clustering.Density;
using AiDotNet.Tensors.LinearAlgebra;

var data = new Matrix<double>(6, 2);
double[][] rows = { new[] { 1.0, 1.0 }, new[] { 1.2, 0.9 }, new[] { 1.1, 1.1 },
                    new[] { 8.0, 8.0 }, new[] { 8.2, 7.9 }, new[] { 7.9, 8.1 } };
for (int i = 0; i < 6; i++) { data[i, 0] = rows[i][0]; data[i, 1] = rows[i][1]; }

var result = await new AiModelBuilder<double, Matrix<double>, Vector<double>>()
    .ConfigureModel(new HDBSCAN<double>())
    .ConfigureDataLoader(DataLoaders.FromMatrix(data))
    .BuildAsync();

var labels = result.Predict(data);
Console.WriteLine($"HDBSCAN: clustered {labels.Length} points.");

Constructors

Constructor Summary
HDBSCAN(HDBSCANOptions<T> options) Initializes a new HDBSCAN instance.

Properties

Property Summary
T[] OutlierScores { get; } Gets the outlier scores for each point.
T[] Probabilities { get; } Gets the cluster membership probabilities.
bool SupportsParameterInitialization { get; }

Methods

Method Summary
IFullModel<T, Matrix<T>, Vector<T>> WithParameters(Vector<T> parameters)
ModelOptions GetOptions()
Vector<T> FitPredict(Matrix<T> x)
Vector<T> Predict(Matrix<T> x)
void Train(Matrix<T> x, Vector<T> y)