An acceleration method for Ten Berge et al.’s algorithm for orthogonal INDSCAL
 Yoshio Takane,
 Kwanghee Jung,
 Heungsun Hwang
 … show all 3 hide
Rent the article at a discount
Rent now* Final gross prices may vary according to local VAT.
Get AccessAbstract
INDSCAL (INdividual Differences SCALing) is a useful technique for investigating both common and unique aspects of K similarity data matrices. The model postulates a common stimulus configuration in a lowdimensional Euclidean space, while representing differences among the K data matrices by differential weighting of dimensions by different data sources. Since Carroll and Chang proposed their algorithm for INDSCAL, several issues have been raised: nonsymmetric solutions, negative saliency weights, and the degeneracy problem. Orthogonal INDSCAL (OINDSCAL) which imposes orthogonality constraints on the matrix of stimulus configuration has been proposed to overcome some of these difficulties. Two algorithms have been proposed for OINDSCAL, one by Ten Berge, Knol, and Kiers, and the other by Trendafilov. In this paper, an acceleration technique called minimal polynomial extrapolation is incorporated in Ten Berge et al.’s algorithm. Simulation studies are conducted to compare the performance of the three algorithms (Ten Berge et al.’s original algorithm, the accelerated algorithm, and Trendafilov’s). Possible extensions of the accelerated algorithm to similar situations are also suggested.
 Arabie, P, Carroll, JD, DeSarbo, WS (1987) Three way scaling and clustering. Sage Publications, Newbury Park
 Bennani Dosse, M, Ten Berge, JMF (2008) The assumption of proportional components when Candecomp is applied to symmetric matrices in the context of Indscal. Psychometrika 73: pp. 303307 CrossRef
 Carroll, JD, Chang, JJ (1970) Individual differences and multidimensional scaling via an Nway generalization of EckartYoung decomposition. Psychometrika 35: pp. 282319 CrossRef
 Carroll JD, Chang JJ (1972) IDIOSCAL (Individual Differences In Orientation SCALing): a generalization of INDSCAL allowing IDIOsyncratic reference systems as well as an analytic approximation to INDSCAL. Paper presented at the Psychometric Society, Princeton
 Carroll, JD, De Soete, G, Pruzansky, S (1989) An evaluation of five algorithms for generating an initial configuration for SINDSCAL. J Classif 6: pp. 105119 CrossRef
 Clarkson, DB (1988) A least squares version of algorithm AS 211: the FG diagonalization algorithm. Appl StatJ Roy St C 37: pp. 317321
 De Leeuw, J, Pruzansky, S. (1978) A new computational method to fit the weighted euclidean distance model. Psychometrika 43: pp. 479490 CrossRef
 Fletcher, R (1981) Practical methods of optimization, constrained optimization, vol. 2. Wiley, Chichester
 Flury, B (1988) Common principal components and related multivariate models. Wiley, New York
 Flury, B, Gautschi, W (1986) An algorithm for simultaneous orthogonal transformation of several positive definite symmetric matrices to nearly diagonal form. SIAM J Sci Stat Comp 7: pp. 169184 CrossRef
 Golub, G, Loan, C (1996) Matrix computations. The Johns Hopkins University Press, London
 Harshman RA (1978) Models for analysis of asymmetrical relationships among N objects or stimuli. Paper presented at the first joint meeting of the psychometric society and the society of mathematical psychology, Hamilton, ON
 Helm, CE (1964) Multidimensional ratio scaling analysis of perceived color relations. J Opt Soc Am 54: pp. 256262 CrossRef
 Jacobowitz D (1975) The acquisition of semantic structures. Unpublished doctoral dissertation, University of North Carolina.
 Jennrich, RI (2001) A simple general procedure for orthogonal rotation. Psychometrika 66: pp. 289306 CrossRef
 Kaiser, HF (1958) The varimax criterion for analytic rotation in factor analysis. Psychometrika 23: pp. 187200 CrossRef
 Kiers, HAL (1989) A computational shortcut for INDSCAL with orthonormality constraints on positive semidefinite matrices of low rank. Comput Stat Quart 2: pp. 119135
 Kiers, HAL (1989) An alternating least squares algorithm for fitting the two and threeway DEDICOM model and the IDIOSCAL model. Psychometrika 54: pp. 515521 CrossRef
 Kiers, HAL (1991) Hierarchical relations among threeway methods. Psychometrika 56: pp. 449470 CrossRef
 Kiers, HAL, Takane, Y (1994) A generalization of GIPSCAL for the analysis of nonsymmetric data. J Classif 11: pp. 7999 CrossRef
 Krijnen, WP, Dijkstra, TK, Stegeman, A (2008) On the nonexistence of optimal solutions and the occurrence of “degeneracy” in the Candecomp/Parafac model. Psychometrika 73: pp. 431439 CrossRef
 Kroonenberg, PM (1983) Threemode principal component analysis. DSWO Press, Leiden
 Lingoes JC, Roskam EE (1973) A mathematical and empirical analysis of two multidimensional scaling algorithms. Psychometrika Monograph Supplement 19
 Loisel, S, Takane, M (2009) Fast indirect robust generalized method of moments. Comput Stat and Data An 53: pp. 35713579 CrossRef
 Nguyen TK, OshimaTakane Y (2008) Do 2 to 3yearold children use functional or shape cues to name objects? Poster presented at the International Society on Infant Studies, Vancouver, BC.
 Rosenberg, S, Kim, MP (1975) The method of sorting as a datagathering procedure in multivariate research. Multivar Behav Res 10: pp. 489502 CrossRef
 Schiffman, SS, Reynolds, ML, Young, FW (1981) Introduction to Multidimensional Scaling. Academic Press, New York
 Smith, DA, Ford, WF, Sidi, A (1987) Extrapolation methods for vector sequences. SIAM Review 29: pp. 199233 CrossRef
 Stegeman, A (2007) Degeneracy in Candecomp/Parafac explained for p × p × 2 arrays of rank p + 1 or higher. Psychometrika 71: pp. 483501 CrossRef
 Takane, Y Applications of multidimensional scaling in psychometrics. In: Rao, CR, Sinharay, S eds. (2007) Handbook of statistics: Pyschometrics, vol 26. Elsevier BV, Amsterdam, pp. 359400
 Takane Y, Zhang Z Algorithms for DEDICOM: acceleration, deceleration, or neither? J Chemometr (in press)
 Ten Berge, JMF (1984) A joint treatment of varimax and the problem of diagonalizing symmetric matrices simultaneously in the least squares sense. Psychometrika 49: pp. 347358 CrossRef
 Ten Berge, JMF, Kiers, HAL (1991) Some clarifiation of the CANDECOMP algorithm applied to INDSCAL. Psychometrika 56: pp. 317326 CrossRef
 Ten Berge, JMF, Kiers, HAL, De Leeuw, J (1988) Explicit candecomp/parafac solutions for a contrived 2 × 2 × 2 array of rank three. Psychometrika 53: pp. 579584 CrossRef
 Ten Berge, JMF, Knol, DL, Kiers, HAL (1988) A treatment of the orthomax rotation family in terms of diagonalization, and a reexamination of a singular value approach to varimax rotation. Comput Stat Quart 3: pp. 207217
 Ten Berge, JMF, Kiers, HAL, Krijnen, WP (1993) Computational solutions for the problem of negative saliences and nonsymmetry in INDSCAL. J Classif 10: pp. 115124 CrossRef
 Trendafilov, N (2002) GIPSCAL revisited: a projected gradient approach. Stat Comput 12: pp. 135145 CrossRef
 Trendafilov, N Orthonormalityconstrained INDSCAL with nonnegative salience. In: Legan, A eds. (2004) Computational science and its applications Lecture Notes in Computer Science 3044 Part II. Springer, Berlin, pp. 952960
 Zhang Z, Takane Y Statistics: multidimensional scaling. In: Baker E, McGaw B, Peterson PP (eds) International encyclopedia of education, 3rd edn. Elsevier, Oxford (in press)
 Title
 An acceleration method for Ten Berge et al.’s algorithm for orthogonal INDSCAL
 Journal

Computational Statistics
Volume 25, Issue 3 , pp 409428
 Cover Date
 20100901
 DOI
 10.1007/s0018001001846
 Print ISSN
 09434062
 Online ISSN
 16139658
 Publisher
 SpringerVerlag
 Additional Links
 Topics
 Keywords

 Multiway data analysis
 Minimal polynomial extrapolation (MPE)
 Singular value decomposition (SVD) algorithm
 Dynamical system algorithm
 15A15MSC code1
 15A21
 15A23
 Industry Sectors
 Authors

 Yoshio Takane ^{(1)}
 Kwanghee Jung ^{(1)}
 Heungsun Hwang ^{(1)}
 Author Affiliations

 1. Department of Psychology, McGill University, 1205 Dr. Penfield Ave, Montreal, QC, H3A 1B1, Canada