Grant: Discovery Grants Program - Individual
McGill University
$15,000.00
Extensions of generalized structured component analysis and regularized fuzzy clusterwise generalizations of statistical methods
EF 2008-09
Résumé
The program of research proposed herein involves two major themes: (1) Extensions of generalized structured component analysis and (2) regularized fuzzy clusterwise generalizations of statistical methods.
(1) Extensions of generalized structured component analysis
Generalized Structured Component Analysis (GSCA) (Hwang & Takane, 2004) was proposed as an alternative to traditional approaches to Structural Equation Modeling (SEM). Although a variety of extensions and refinements of GSCA have been proposed over the past few years, its analytic capability and generality need to be further improved so as to become a mainstream approach to SEM. Thus, the first theme of the proposed program involves rendering GSCA capable of effectively addressing many advanced SEM issues including nonlinear latent variables, endogenous higher-order latent variables, ridge estimation, and maximum-likelihood estimation.
(2) Regularized fuzzy clusterwise generalizations of statistical methods
Many statistical methods estimate parameters by pooling the data across observations under the implicit assumption that all observations are sampled from a single homogenous population. Hence, they are not ideally suited for investigating the existence of subgroups of observations which exhibit distinctive response patterns. The second theme of the proposed program aims to generalize three popular statistical methods (generalized linear models, partial least squares regression, and multiple correspondence analysis) so as to take into account such group-level heterogeneity. Each of these methods is combined with regularized fuzzy clustering in a unified framework for capturing this heterogeneity.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
311881-2008
Name nom
Hwang, Heungsun
Department département
Psychology
Organization id
61
Institution établissement
McGill University
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2008
Competition year année de concours
2008
Award amount
15000
Program id
RGPIN
Program nane en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
Discovery grants programs
Group fr
Programmes de subventions à la découverte
Committee code
14
Committee name en
Statistical Sciences
Committee name fr
Statistique
Area of application code
1210
Area of application group en
Advancement of knowledge
Area of application group fr
Avancement des connaissances
Area of application en
Psychology
Area of application fr
Psychologie
Research subject code
3011
Research subject group en
Statistics and probability
Research subject group fr
Statistique et probabilité
Research subject en
Multivariate analysis
Research subject fr
Analyse multivariée
Partie
1
Nb partie
5
Application title
Extensions of generalized structured component analysis and regularized fuzzy clusterwise generalizations of statistical methods
Keyword
NA
Application summary
The program of research proposed herein involves two major themes: (1) Extensions of generalized structured component analysis and (2) regularized fuzzy clusterwise generalizations of statistical methods.
(1) Extensions of generalized structured component analysis
Generalized Structured Component Analysis (GSCA) (Hwang & Takane, 2004) was proposed as an alternative to traditional approaches to Structural Equation Modeling (SEM). Although a variety of extensions and refinements of GSCA have been proposed over the past few years, its analytic capability and generality need to be further improved so as to become a mainstream approach to SEM. Thus, the first theme of the proposed program involves rendering GSCA capable of effectively addressing many advanced SEM issues including nonlinear latent variables, endogenous higher-order latent variables, ridge estimation, and maximum-likelihood estimation.
(2) Regularized fuzzy clusterwise generalizations of statistical methods
Many statistical methods estimate parameters by pooling the data across observations under the implicit assumption that all observations are sampled from a single homogenous population. Hence, they are not ideally suited for investigating the existence of subgroups of observations which exhibit distinctive response patterns. The second theme of the proposed program aims to generalize three popular statistical methods (generalized linear models, partial least squares regression, and multiple correspondence analysis) so as to take into account such group-level heterogeneity. Each of these methods is combined with regularized fuzzy clustering in a unified framework for capturing this heterogeneity.