The weighted Euclidean model (INDSCAL) is the most well-known and used model of multidimensional scaling of three-way data. INDSCAL states a unique representation for the objects (common configuration space) and for each occasion weights for the dimensions of this representation (individual differences weights), thus definitely assuming that there are not systematic ``strong'' differences between data dissimilarity sources. Otherwise, when heterogeneous occasions are observed, it is shown that INDSCAL may fail to identify a common space representative of the observed data structure. The heterogeneous INDSCAL in K classes model, simply called K-INDSCAL, is proposed to handle the heterogeneity in the data. The model includes the individual weights in order to preserve the rotational invariance of the INDSCAL model. The parameters of the model are estimated in a least-squares fitting context and an efficiently coordinate descent algorithm is given.
The weighted Euclidean model (INDSCAL) is the most well-known and used model of multidimensional scaling of three-way data. INDSCAL states a unique representation for the objects (common configuration space) and for each occasion weights for the dimensions of this representation (individual differences weights), thus definitely assuming that there are not systematic ``strong'' differences between data dissimilarity sources. Otherwise, when heterogeneous occasions are observed, it is shown that INDSCAL may fail to identify a common space representative of the observed data structure. The heterogeneous INDSCAL in K classes model, simply called K-INDSCAL, is proposed to handle the heterogeneity in the data. The model includes the individual weights in order to preserve the rotational invariance of the INDSCAL model. The parameters of the model are estimated in a least-squares fitting context and an efficiently coordinate descent algorithm is given.
The K-INDSCAL Model for Heterogeneous Three-Way Dissimilarity Data / Bocci, Laura; Vichi, Maurizio. - STAMPA. - (2008), pp. 17-17. (Intervento presentato al convegno 32nd Annual Conference on Advances in Data Analysis, Data Handling and Business Intelligence tenutosi a Hamburg nel July 16-18, 2008).
The K-INDSCAL Model for Heterogeneous Three-Way Dissimilarity Data
BOCCI, Laura;VICHI, Maurizio
2008
Abstract
The weighted Euclidean model (INDSCAL) is the most well-known and used model of multidimensional scaling of three-way data. INDSCAL states a unique representation for the objects (common configuration space) and for each occasion weights for the dimensions of this representation (individual differences weights), thus definitely assuming that there are not systematic ``strong'' differences between data dissimilarity sources. Otherwise, when heterogeneous occasions are observed, it is shown that INDSCAL may fail to identify a common space representative of the observed data structure. The heterogeneous INDSCAL in K classes model, simply called K-INDSCAL, is proposed to handle the heterogeneity in the data. The model includes the individual weights in order to preserve the rotational invariance of the INDSCAL model. The parameters of the model are estimated in a least-squares fitting context and an efficiently coordinate descent algorithm is given.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
