scientific article; zbMATH DE number 1107548
This chapter describes a model, and corresponding algorithms, for two-way two-mode matrices with binary \((0,1)\) cell entries. In the model, the objects and the attributes are each subdivided into a number of clusters, called object classes and attribute classes, respectively. The model is a comprehensive representation of the structural relations among object and attribute classes, as well as the association between the object and attribute structures. It is closely related to blockmodeling, to Boolean factor analysis, and to additive clustering. The present model is similar to blockmodeling in that equivalent objects and equivalent attributes are grouped, respectively, into classes. The algorithm is similar to Boolean factor analysis in that it is based on a Boolean decomposition of the two-way data matrix. Additive clustering is included here since its underlying model depends heavily on parameters constrained to be binary. However, instead of attributes, hypothetical variables called ``features are assumed to be the basis for the observed similarities among the objects. An important feature of the present model is that it postulates an order relation among the object classes and among the attribute classes, using a set-theoretical framework. This set-theoretical formulation implies a structure that is different from those of Boolean factor analysis, blockmodeling, and additive clustering. The set-theoretical framework is especially suited to binary data. We describe first the basic characteristics of the set-theoretical model and its associated algorithm, HICLAS. A hypothetical example from person perception is used to explicate both the set-theoretical features implicit in a two-way binary matrix and the kind of substantive information that results from the application of the HICLAS algorithm to such data. What then follows is a description of how external properties of objects and/or attributes can be projected into a set-theoretical structure. The hypothetical example from person perception is again used to illustrate this projection method and demonstrate the flexibility it adds to a HICLAS analysis and, hence, to substantive application. The remaining part of the chapter is a summary of selected empirical applications of this structural model and of the projection of external properties. For the entire collection see [Zbl 0836.00014].
- scientific article; zbMATH DE number 1113986
- Hierarchical classes: model and data analysis
- Models for ordinal hierarchical classes analysis
- Hierarchical graphical Bayesian models in psychology
- scientific article; zbMATH DE number 5280185
- scientific article; zbMATH DE number 475231
- scientific article; zbMATH DE number 1405304
- Hierarchical latent class models for cluster analysis
- Hierarchical diagnostic classification models: a family of models for estimating and testing attribute hierarchies
- An evaluation of two algorithms for hierarchical classes analysis.
- Properties of ideal point classification models for bivariate binary data
- INDCLAS: a three-way hierarchical classes model
- A variable neighborhood search method for generalized blockmodeling of two-mode binary matrices
- Models for ordinal hierarchical classes analysis
- Hierarchical classes: model and data analysis
- Design and implementation of a web mining system for organizing search engine results
- Probability matrix decomposition models
- The conjunctive model of hierarchical classes
- The SIMCLAS model: simultaneous analysis of coupled binary data matrices with noise heterogeneity between and within data blocks
- Inducing a blockmodel structure of two-mode binary data using seriation procedures
- CLASSI: A classification model for the study of sequential processes and individual differences therein
- Uniqueness of real-valued hierarchical classes models
This page was built for publication:
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4373477)