Abstract
In the context of three-way proximity data, an INDCLUS-type model is presented to address the issue of subject heterogeneity regarding the perception of object pairwise similarity. A model, termed ROOTCLUS, is presented that allows for the detection of a subset of objects whose similarities are described in terms of non-overlapping clusters (ROOT CLUSters) common across all subjects. For the other objects, Individual partitions, which are subject specific, are allowed where clusters are linked one-to-one to the Root clusters. A sound ALS-type algorithm to fit the model to data is presented. The novel method is evaluated in an extensive simulation study and illustrated with empirical data sets.
Similar content being viewed by others
Notes
The relative loss is defined here as the ratio of the raw loss to the total sum of squares of the data.
The MATLAB code is available online on SpringerLink with the article.
Given two matrices \(\mathbf {A}\) and \(\mathbf {B}\) with the same number J of columns, the Khatri–Rao product of \(\mathbf {A}\) and \(\mathbf {B}\) is the column-wise Kronecker product, i.e., \(\mathbf {A} |\otimes | \mathbf {B} = (\mathbf {a}_1 \otimes \mathbf {b}_1, \ldots , \mathbf {a}_j \otimes \mathbf {b}_j, \ldots , \mathbf {a}_J \otimes \mathbf {b}_J )\) where \(\mathbf {a}_j\) and \(\mathbf {b}_j\) are the j-th (\(j=1,\ldots ,J\)) column of \(\mathbf {A}\) and \(\mathbf {B}\), respectively, and \(\otimes \) denotes the Kronecker product.
The Kappa coefficient (KC) between two binary matrices is equal to the proportion of agreement between the two matrices (i.e., the proportion of the corresponding cells having the same values), corrected for chance (Wilderjans et al. 2012):
$$\begin{aligned} KC=\frac{(p_{00} + p_{11}) - (p_{0.}p_{.0} + p_{1.}p_{.1})}{1 - (p_{0.}p_{.0} + p_{1.}p_{.1})}, \end{aligned}$$with \(p_{00}\)\((p_{11})\) being the proportion of corresponding cells that both are zero (one) and \(p_{0.}\) and \(p_{1.}\) (\(p_{.0}\) and \(p_{.1}\)) the marginal proportion of zero- and one-cells for the first (second) matrix. Note that \(p_{00}+p_{11}\) equals the (uncorrected) proportion of corresponding cells that have the same value.
Note that \(w_2\) is missing here because the Root cluster \(R_2\) is a singleton and the diagonal entries of the similarity matrices are not fitted in this application.
References
Bocci, L., & Vicari, D. (2017). GINDCLUS: Generalized INDCLUS with external information. Psychometrika, 82, 355–381.
Bocci, L., Vicari, D., & Vichi, M. (2006). A mixture model for the classification of three-way proximity data. Computational Statistics & Data Analysis, 50, 1625–1654.
Calinski, T., & Harabasz, J. (1974). A dendrite method for cluster analysis. Communications in Statistics, 3, 1–27.
Carroll, J. D., & Arabie, P. (1983). INDCLUS: An individual differences generalization of ADCLUS model and the MAPCLUS algorithm. Psychometrika, 48, 157–169.
Carroll, J. D., & Chang, J. J. (1970). Analysis of individual differences in multidimensional scaling via an N-generalization of the Eckart–Young decomposition. Psychometrika, 35, 283–319.
Chaturvedi, A., & Carroll, J. D. (2006). CLUSCALE (CLUstering and multidimensional SCAL[E]ing): A three-way hybrid model incorporating clustering and multidimensional scaling structure. Journal of Classification, 23, 269–299.
Chaturvedi, A. J., & Carroll, J. D. (1994). An alternating combinatorial optimization approach to fitting the INDCLUS and generalized INDCLUS models. Journal of Classification, 11, 155–170.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Measurement, 20, 37–46.
De Leeuw, J. (1994). Block-relaxation algorithms in statistics. In H. H. Bock, W. Lenski, & M. M. Richter (Eds.), Information systems and data analysis (pp. 308–325). Berlin: Springer.
Giordani, P., & Kiers, H. A. L. (2012). FINDCLUS: Fuzzy INdividual Differences CLUStering. Journal of Classification, 29, 170–198.
Gordon, A. D., & Vichi, M. (1998). Partitions of Partitions. Journal of Classification, 15, 265–285.
Hubert, L. J., & Arabie, P. (1985). Comparing partitions. Journal of Classification, 2, 193–218.
Hubert, L. J., Arabie, P., & Meulman, J. (2006). The structural representation of proximity matrices with MATLAB. Philadelphia: SIAM.
Kiers, H. A. L. (1997). A modification of the SINDCLUS algorithm for fitting the ADCLUS and INDCLUS models. Journal of Classification, 14, 297–310.
Lawson, C. L., & Hanson, R. J. (1974). Solving least squares problems. Englewood Cliffs: Prentice Hall.
McDonald, R. P. (1980). A simple comprehensive model for the analysis of covariance structures: Some remarks on applications. British Journal of Mathematical and Statistical Psychology, 33, 161–183.
Mirkin, B. G. (1987). Additive clustering and qualitative factor analysis methods for similarity matrices. Journal of Classification, 4, 7–31.
Rao, C. R., & Mitra, S. (1971). Generalized inverse of matrices and its applications. New York: Wiley.
Rocci, R., & Vichi, M. (2008). Two-mode multi-partitioning. Computational Statistics & Data Analysis, 52, 1984–2003.
Shepard, R. N., & Arabie, P. (1979). Additive clustering: Representation of similarities as combinations of discrete overlapping properties. Psychological Review, 86, 87–123.
Schepers, J., Ceulemans, E., & Van Mechelen, I. (2008). Selecting among multi-mode partitioning models of different complexities: A comparison of four model selection criteria. Journal of Classification, 25, 67–85.
Schiffman, S. S., Reynolds, M. L., & Young, F. W. (1981). Introduction to multidimensional scaling. London: Academic Press.
Vicari, D., & Vichi, M. (2009). Structural classification analysis of three-way dissimilarity data. Journal of Classification, 26, 121–154.
Vichi, M. (1999). One mode classification of a three-way data set. Journal of Classification, 16, 27–44.
Wedel, M., & DeSarbo, W. S. (1998). Mixtures of (constrained) ultrametric trees. Psychometrika, 63, 419–443.
Wilderjans, T. F., Depril, D., & Van Mechelen, I. (2012). Block-relaxation approaches for fitting the INDCLUS model. Journal of Classification, 29, 277–296.
Acknowledgements
The authors are grateful to the Associate Editor and referees for their valuable comments and suggestions which greatly improved the presentation and content of the first version.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendix
Appendix
In order to solve the constrained problem (14) in those cases when the diagonal entries of the H similarity matrices \(\mathbf {S}_{h}\) are not of interest, the steps 1 to 4 of the ALS-type algorithm presented in Sect. 4 can be modified straightforwardly as follows.
Since only the off-diagonal elements of matrices \(\mathbf {S}_{h}\) (\(h=1,\ldots ,H\)) need to be considered, the loss function (14) becomes
where
and \(\left\| \mathbf {Z} \right\| _{\text {off}}^2 = \sum _{x=1}^{X}\sum _{\begin{array}{c} y=1\; ;\; \end{array}{y\ne x}}^{Y} z_{xy}^2\).
In step 1, the loss function (23), instead of (14), is minimized over \(\mathbf {P}\) and \(\mathbf {M}_h\) (\(h=1,\ldots ,H\)).
In steps 2 to 4, all the rows of \(\mathbf {s}_{h}\), \(\mathbf {T}\), \(\mathbf {Q}_{h}\) and \(\mathbf {1}_{N^2}\) in model (15), corresponding to the diagonal entries of the matrices in (14), need to be left out. Such reduced structures are obtained as follows:
where \(\odot \) denotes the Hadamard product, \(\mathbf {d}\) is the column vector of size \(N^2\) of the vectorized matrix \(\big (\mathbf {1}_N \mathbf {1}_N^\prime - \mathbf {I}_N\big )\), being \(\mathbf {I}_N\) the identity matrix of size N, and \(\mathbf {D}\) is the \(N^2 \times J\) matrix having all its columns equal to \(\mathbf {d}\).
Therefore, model (15) is rewritten in terms of (25)–(28) and Steps 2, 3 and 4 modified accordingly.
Rights and permissions
About this article
Cite this article
Bocci, L., Vicari, D. ROOTCLUS: Searching for “ROOT CLUSters” in Three-Way Proximity Data. Psychometrika 84, 941–985 (2019). https://doi.org/10.1007/s11336-019-09686-1
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11336-019-09686-1