Click or drag to resize

ModifiedGammaStatisticTInstance Class

Implements an internal evaluation method based on a modified/improved version of Hubert's Gamma (Γ) statistic in [1] with the transformation introduced in [2] in order to be maximized.
Inheritance Hierarchy
SystemObject
  Aglomera.Evaluation.InternalModifiedGammaStatisticTInstance

Namespace:  Aglomera.Evaluation.Internal
Assembly:  Aglomera (in Aglomera.dll) Version: 1.1.0
Syntax
C#
Copy
public class ModifiedGammaStatistic<TInstance> : IInternalEvaluationCriterion<TInstance>
where TInstance : Object, IComparable<TInstance>
Request Example View Source

Type Parameters

TInstance
The type of instance considered.

The ModifiedGammaStatisticTInstance type exposes the following members.

Constructors
  NameDescription
Public methodModifiedGammaStatisticTInstance
Creates a new ModifiedGammaStatisticTInstance with given dissimilarity metric.
Top
Properties
  NameDescription
Public propertyDissimilarityMetric
Gets the metric used by this criterion to measure the dissimilarity / distance between cluster elements.
Top
Methods
  NameDescription
Public methodEquals (Inherited from Object.)
Public methodEvaluate
Evaluates the given ClusterSetTInstance partition according to this evaluation criterion.
Protected methodFinalize (Inherited from Object.)
Public methodGetHashCode (Inherited from Object.)
Public methodGetType (Inherited from Object.)
Protected methodMemberwiseClone (Inherited from Object.)
Public methodToString (Inherited from Object.)
Top
Remarks
Notes: - "The improved Hubert Γ statistic describes the degree of a partition fitting the data set. [...] The partition number fitting data set may be discovered from the plot of Γ versus n, however, it is difficult to find the inflexion from the plot and it is possible that the partition number obtained is just close to the best solution, but not that we want. So, it is not feasible in practice to determine the optimal partition by the plot of Γ versus n directly. [...] In the plot of [transformed Γ] versus [number of clusters c], Γ goes to zero with c close to n, and a max peak value that corresponds to a significant increase of c can be found. The number of clusters at which the peak value occurs is equal to the number of clusters fitting the data." [2] References: [1] - Hubert, L., & Arabie, P. (1985). Comparing partitions. Journal of classification, 2(1), 193-218. [2] - Zhao, H., Liang, J., & Hu, H. (2006, August). Clustering Validity Based on the Improved Hubert\Gamma Statistic and the Separation of Clusters. In First International Conference on Innovative Computing, Information and Control, 2006. ICICIC'06. (Vol. 2, pp. 539-543). IEEE.
See Also