BisectingKMeansOptions<T>
Options & Configuration · AiDotNet.Clustering.Options
Configuration options for Bisecting K-Means clustering.
Inheritance: ModelOptions → ClusteringOptions<T> → BisectingKMeansOptions<T>
For Beginners
Bisecting K-Means works by:
- Starting with all data in one big cluster
- Splitting the largest cluster into two using K-Means
- Repeating step 2 until you have k clusters
Advantages over regular K-Means:
- Often produces better clusters (more balanced)
- Less sensitive to initialization
- Builds a cluster hierarchy as a side effect
How It Works
Bisecting K-Means is a divisive hierarchical clustering algorithm that starts with all points in one cluster and recursively bisects clusters until the desired number of clusters is reached.
Constructors
| Constructor | Summary |
|---|---|
BisectingKMeansOptions() |
Properties
| Property | Summary |
|---|---|
BisectionClusterSelection ClusterSelection { get; set; } |
Gets or sets the cluster selection method for bisection. |
bool BuildHierarchy { get; set; } |
Gets or sets whether to build a hierarchy tree during clustering. |
int MinClusterSizeForBisection { get; set; } |
Gets or sets the minimum cluster size that can be bisected. |
int NumBisectionTrials { get; set; } |
Gets or sets the number of bisection trials at each split. |
int NumClusters { get; set; } |
Gets or sets the number of clusters to find. |