Enumerates the methods that may be used to initialize
the K-means clustering algorithm.
Namespace:
Extreme.Statistics.Multivariate
Assembly:
Extreme.Numerics (in Extreme.Numerics.dll) Version: 8.1.1
public enum KMeansInitializationMethod
Public Enumeration KMeansInitializationMethod
public enum class KMeansInitializationMethod
type KMeansInitializationMethod
| Member name | Value | Description |
---|
| KMeansPlusPlus | 0 |
Use the K-means++ algorithm to compute initial centroids.
This is the default.
|
| RandomCenters | 1 |
Use randomly selected observations as the centroid.
|
| Forgy | 1 |
Same as RandomCenters.
Use randomly selected observations as the centroid.
|
| RandomAssignments | 2 |
Assign each observation randomly to one of the clusters and uses the centroid
of each cluster.
|
Reference