salso (version 0.1.16)

confidence: Compute Clustering Confidence

Description

This function computes the confidence values for n observations based on a clustering estimate and the pairwise similarity matrix.

Usage

confidence(estimate, psm)

Arguments

estimate

A vector of length n, where i and j are in the same subset (i.e., cluster) if and only if estimate[i] == estimate[j].

psm

A pairwise similarity matrix, i.e., n-by-n symmetric matrix whose (i,j) element gives the (estimated) probability that items i and j are in the same subset (i.e., cluster) of a partition (i.e., clustering).

Value

A list of the following elements:

estimate

The value of the estimate argument.

psm

The value of the psm argument.

confidence

A numeric vector with the same length as estimate that contains the mean probability that each item is paired with all of the other items in its subset (i.e., cluster).

confidenceMatrix

A matrix containing the mean confidences of items in each subset on the diagonal. In the off-diagonal elements, the mean confidence among all pairs from the two subsets. High probabilities on the diagonal and low probabilities everywhere else indicate good separability among the subsets.

exemplar

A numeric vector containing the exemplar for each subset (i.e, cluster). The "exemplar" of a subset has the highest probability of being clustered with all of the other items in its subset.

order

A vector giving the permutation of the original observations used in plotting.

See Also

plot.salso.confidence, salso, dlso, psm

Examples

# NOT RUN {
# Use 'parallel=FALSE' per CRAN rules for examples but, in practice, omit this.
probs <- psm(iris.clusterings, parallel=FALSE)
est <- salso(probs, parallel=FALSE)$estimate
conf <- confidence(est, probs)
conf

# }