nomclust (version 0.91.1010)

eskin: Eskin Measure

Description

The Eskin similarity measure was proposed by Eskin et al. (2002). It is constructed to assign higher weights to mismatches on variables with more categories, see (Boriah et al., 2008). Hierarchical clustering methods require a proximity (dissimilarity) matrix instead of a similarity matrix as an entry for the analysis; therefore, dissimilarity D is computed from similarity S according the equation 1/S-1. The use and evaluation of clustering with this measure can be found e.g. in (Sulc and Rezankova, 2014).

Usage

eskin(data)

Arguments

data

data frame with cases in rows and variables in colums. Cases are characterized by nominal (categorical) variables coded as numbers.

Value

• Function returns a matrix of the size n x n, where n is a number of objects in original data. The matrix contains proximities between all pairs of objects. It can be used in hierarchical cluster analyses (HCA), e.g. in agnes.

References

Boriah, S., Chandola and V., Kumar, V. (2008). Similarity measures for categorical data: A comparative evaluation. In: Proceedings of the 8th SIAM International Conference on Data Mining, SIAM, p. 243-254. Available at: http://www-users.cs.umn.edu/~sboriah/PDFs/BoriahBCK2008.pdf. Eskin, E., Arnold, A., Prerau, M., Portnoy, L. and Stolfo, S. (2002). A geometric framework for unsupervised anomaly detection. In D. Barbara and S. Jajodia (Eds): Applications of Data Mining in Computer Security, p. 78-100. Norwell: Kluwer Academic Publishers. Sulc, Z. and Rezankova, H. (2014). Evaluation of recent similarity measures for categorical data. In: AMSE. Wroclaw: Wydawnictwo Uniwersytetu Ekonomicznego we Wroclawiu, p. 249-258. Available at: http://www.amse.ue.wroc.pl/papers/Sulc,Rezankova.pdf.

See Also

good1, good2, good3, good4, iof, lin, lin1, morlini, of, sm.

Examples

#sample data
data(data20)
# Creation of proximity matrix
prox_eskin <- eskin(data20)