CUREOptions<T>
Options & Configuration · AiDotNet.Clustering.Options
Configuration options for CURE clustering algorithm.
Inheritance: ModelOptions → ClusteringOptions<T> → CUREOptions<T>
For Beginners
CURE is designed to find non-spherical clusters:
- Uses multiple “representative” points per cluster instead of just a center
- These points are pulled slightly toward the center (shrinking)
- This helps find oddly-shaped clusters (like bananas or spirals)
It’s like describing a cluster by several key locations within it, rather than just one center point.
How It Works
CURE (Clustering Using REpresentatives) is a hierarchical clustering algorithm that represents each cluster by a set of well-scattered representative points, which are then shrunk toward the cluster center.
Constructors
| Constructor | Summary |
|---|---|
CUREOptions() |
Properties
| Property | Summary |
|---|---|
bool UsePartitioning { get; set; } |
Gets or sets whether to use random partitioning. |
double SampleFraction { get; set; } |
Gets or sets the sample fraction for large datasets. |
double ShrinkFactor { get; set; } |
Gets or sets the shrink factor for representative points. |
int NumClusters { get; set; } |
Gets or sets the number of clusters to find. |
int NumPartitions { get; set; } |
Gets or sets the number of partitions for parallel processing. |
int NumRepresentatives { get; set; } |
Gets or sets the number of representative points per cluster. |