Three-mode partitioning. (English) Zbl 1157.62447

Summary: The three-mode partitioning model is a clustering model for three-way three-mode data sets that implies a simultaneous partitioning of all three modes involved in the data. In the associated data analysis, a data array is approximated by a model array that can be represented by a three-mode partitioning model of a prespecified rank, minimizing a least squares loss function in terms of differences between data and model. Algorithms have been proposed for this minimization, but their performance is not yet clear. A framework for alternating least-squares methods is described in order to offset the performance problem. Furthermore, a number of both existing and novel algorithms are discussed within this framework. An extensive simulation study is reported in which these algorithms are evaluated and compared according to sensitivity to local optima. The recovery of the truth underlying the data is investigated in order to assess the optimal estimates. The ordering of the algorithms with respect to performance in finding the optimal solution appears to change as compared to the results obtained from the simulation study when a collection of four empirical data sets have been used. This finding is attributed to violations of the implicit stochastic model underlying both the least-squares loss function and the simulation study. Support for the latter attribution is found in a second simulation study.


62H30 Classification and discrimination; cluster analysis (statistical aspects)
65C60 Computational problems in statistics (MSC2010)
Full Text: DOI


[1] Al-Sultan, K.S.; Maroof Khan, M., Computational experience on four algorithms for the hard clustering problem, Pattern recognition lett., 17, 295-308, (1996)
[2] Baier, D.; Gaul, W.; Schader, M., Two-mode overlapping clustering with applications to simultaneous benefit segmentation and market structuring, (), 557-566
[3] Banfield, J.D.; Raftery, A.E., Model-based gaussian and non-Gaussian clustering, Biometrics, 49, 803-821, (1993) · Zbl 0794.62034
[4] Bock, H.H., Probabilistic models in cluster analysis, Comput. statist. data anal., 23, 5-28, (1996) · Zbl 0900.62324
[5] Bryant, P.; Williamson, J.A., Asymptotic behaviour of classification maximum likelihood estimates, Biometrika, 65, 273-281, (1978) · Zbl 0393.62011
[6] Carroll, J.D.; Arabie, P., Multidimensional scaling, Annu. rev. psychol., 31, 607-649, (1980)
[7] Celeux, G.; Govaert, G., Comparison of the mixture and the classification maximum likelihood in cluster analysis, J. statist. comput. simulation, 14, 315-332, (1992) · Zbl 0937.62605
[8] Chaturvedi, A.; Carroll, J.D., An alternating combinatorial optimization approach to Fitting the {\scindclus} and generalized {\scindclus} models, J. classification, 11, 155-170, (1994) · Zbl 0825.62539
[9] Gaul, W.; Schader, M., A new algorithm for two-mode clustering, (), 15-23 · Zbl 0896.62060
[10] Govaert, G., Simultaneous clustering of rows and columns, Control cybernet., 24, 437-458, (1995) · Zbl 0852.62055
[11] Govaert, G.; Nadif, M., Clustering with block mixture models, Pattern recognition, 36, 463-473, (2003)
[12] Hand, D.J.; Krzanowski, W.J., Optimising k-means clustering results with standard software packages, Comput. statist. data anal., 49, 969-973, (2005) · Zbl 1429.62244
[13] Hubert, L.; Arabie, P., Comparing partitions, J. classification, 2, 193-218, (1985)
[14] Hubert, L.; Arabie, P.; Meulman, J., Combinatorial data analysis: optimization by dynamic programming, (2001), SIAM Philadelphia, PA · Zbl 0999.90047
[15] John, S., On identifying the population of origin of each observation in a mixture of observations from two normal populations, Technometrics, 12, 553-563, (1970)
[16] Kiers, H.A.L., 2004. Clustering all three modes of three-mode data: computational possibilities and problems, COMPSTAT 2004, Charles University, Prague. · Zbl 1170.62340
[17] Kroonenberg, P.M.; De Leeuw, J., Principal component analysis of three-mode data by means of alternating least squares algorithms, Psychometrika, 45, 69-97, (1980) · Zbl 0431.62035
[18] Kuppens, P., Van Mechelen, I., in press. Determinants of the anger appraisals of threatened self-esteem, other-blame, and frustration. Cognition Emotion.
[19] McLachlan, G., The classification and mixture maximum likelihood approaches to cluster analysis, (), 199-208
[20] Mezzich, J.E.; Solomon, H., Taxonomy and behavioral science: comparative performance of grouping methods, (1980), Academic Press London
[21] Milligan, G.W., An examination of the effect of six types of error perturbation on fifteen clustering algorithms, Psychometrika, 45, 325-342, (1980)
[22] Murakami, T.; Kroonenberg, P.M., Three-mode models and individual differences in semantic differential data, Multivariate behav. res., 38, 247-283, (2003)
[23] Murtagh, F., (review of the book data, expert knowledge and decisions), J. classification, 6, 129-132, (1989)
[24] Paterlini, S.; Krink, T., Differential evolution and particle swarm optimisation in partitional clustering, Comput. statist. data anal., 50, 1220-1247, (2006) · Zbl 1431.62268
[25] Rocci, R.; Vichi, M., Three-mode clustering of a three-way data set, CLADAG 2003, (2003), University of Bologna Bologna
[26] Schepers, J.; Van Mechelen, I., Three-mode partitioning: model and algorithm, gfkl 2004, (2004), University of Dortmund Dortmund
[27] Selim, S.; Ismail, M., K-means type algorithms: a generalized convergence theorem and characterization of local optimality, IEEE trans. pattern anal. Mach. intell., 6, 81-87, (1984) · Zbl 0546.62037
[28] Steinley, D., Local optima in K-means clustering: what you don’t know may hurt you, Psychol. methods, 8, 294-304, (2003)
[29] Van Coillie, H.; Van Mechelen, I., Expected consequences of anger-related behaviors, European J. personality, 20, 137-154, (2006)
[30] Van Mechelen, I.; Bock, H.-H.; De Boeck, P., Two-mode clustering methods: a structured overview, Statist. methods med. res., 13, 363-394, (2004) · Zbl 1053.62078
[31] Vichi, M., Double k-means clustering for simultaneous classification of objects and variables, (), 43-51
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.