SeededKMeansOptions<T>
Options & Configuration · AiDotNet.Clustering.Options
Configuration options for Seeded K-Means.
Inheritance: ModelOptions → ClusteringOptions<T> → SeededKMeansOptions<T>
For Beginners
Instead of random initialization, use data points you already know the labels for.
Example:
- You have 1000 customer profiles
- You’ve manually labeled 50 as “budget”, “premium”, or “enterprise”
- Seeded K-Means uses those 50 to initialize 3 clusters
- Then it clusters the remaining 950 automatically
This often produces better results than random initialization.
How It Works
Seeded K-Means uses pre-labeled data points as initial cluster seeds. The algorithm starts with these seeds and then proceeds like regular K-Means.
Constructors
| Constructor | Summary |
|---|---|
SeededKMeansOptions() |
Initializes SeededKMeansOptions with appropriate defaults. |
Properties
| Property | Summary |
|---|---|
Dictionary<int, int> Seeds { get; set; } |
Gets or sets the labeled seed points. |
bool ConstrainSeeds { get; set; } |
Gets or sets whether seeded points are constrained to their original cluster. |
int NumClusters { get; set; } |
Gets or sets the number of clusters (inferred from seeds if not set). |