Categories &

Functions List

Class Definition: ClusterCriterion

statistics: obj = ClusterCriterion (x, clust, criterion)

A clustering evaluation object as created by evalclusters.

ClusterCriterion is a superclass for clustering evaluation objects as created by evalclusters.

List of public properties:

ClusteringFunction

a valid clustering function name or function handle. It can be empty if the clustering solutions are passed as an input matrix.

CriterionName

a valid criterion name to evaluate the clustering solutions.

CriterionValues

a vector of values as generated by the evaluation criterion for each clustering solution.

InspectedK

the list of proposed cluster numbers.

Missing

a logical vector of missing observations. When there are NaN values in the data matrix, the corresponding observation is excluded.

NumObservations

the number of non-missing observations in the data matrix.

OptimalK

the optimal number of clusters.

OptimalY

the clustering solution corresponding to OptimalK.

X

the data matrix.

List of public methods:

addK

add a list of numbers of clusters to evaluate.

compact

return a compact clustering evaluation object. Not implemented

plot

plot the clustering evaluation values against the corresponding number of clusters.

See also: evalclusters, CalinskiHarabaszEvaluation, DaviesBouldinEvaluation, GapEvaluation, SilhouetteEvaluation

Source Code: ClusterCriterion

Properties

ClusterCriterion.ClusteringFunction is not documented.
ClusterCriterion.CriterionName is not documented.
ClusterCriterion.CriterionValues is not documented.
ClusterCriterion.InspectedK is not documented.
ClusterCriterion.Missing is not documented.
ClusterCriterion.NumObservations is not documented.
ClusterCriterion.OptimalK is not documented.
ClusterCriterion.OptimalY is not documented.
ClusterCriterion.X is not documented.

Methods

ClusterCriterion: obj = addK (obj, K)

Add a new cluster array to inspect the ClusterCriterion object.

ClusterCriterion: plot (obj)
ClusterCriterion: h = plot (obj)

Plot the evaluation results.

Plot the CriterionValues against InspectedK from the ClusterCriterion, obj, to the current plot. It can also return a handle to the current plot.

ClusterCriterion: obj = compact (obj)

Return a compact ClusterCriterion object (not implemented yet).