 | |
DunnIndexTInstance Class | |
Implements the internal evaluation method in [1] that measures the ratio between the smallest distance between
observations not in the same cluster to the largest intra-cluster distance. The Dunn Index has a value between zero
and infinity, and a higher index indicates a better clustering.
The aim is to identify sets of clusters that are compact, with a small variance between members of the cluster, and
well separated, where the means of different clusters are sufficiently far apart, as compared to the within cluster
variance [2].
Inheritance HierarchyNamespace: Aglomera.Evaluation.Internal
Assembly: Aglomera (in Aglomera.dll) Version: 1.1.0
SyntaxC#
public class DunnIndex<TInstance> : IInternalEvaluationCriterion<TInstance> where TInstance : Object, IComparable<TInstance>
Type Parameters
- TInstance
- The type of instance considered.
The DunnIndexTInstance type exposes the following members.
Constructors| Name | Description | |
|---|---|---|
 | DunnIndexTInstance |
Creates a new DunnIndexTInstance with given dissimilarity metric.
|
Properties| Name | Description | |
|---|---|---|
 | DissimilarityMetric |
Gets the metric used by this criterion to measure the dissimilarity / distance between cluster elements.
|
Methods| Name | Description | |
|---|---|---|
| Equals | (Inherited from Object.) |
| Evaluate |
Evaluates the given ClusterSetTInstance partition according to this evaluation criterion.
|
 | Finalize | (Inherited from Object.) |
| GetHashCode | (Inherited from Object.) |
| GetType | (Inherited from Object.) |
| MemberwiseClone | (Inherited from Object.) |
| ToString | (Inherited from Object.) |
Remarks
Notes:
- This formulation has a peculiar problem, in that if one of the clusters is badly behaved, where the others are
tightly packed, since the denominator contains a 'max' term instead of an average term, the Dunn Index for that set
of clusters will be uncharacteristically low [2].
References:
[1] -
Dunn, J. C. (1973). A fuzzy relative of the ISODATA process and its use in detecting compact well-separated
clusters.
[2] - https://en.wikipedia.org/wiki/Dunn_index
See Also