Strukturanalyse unscharfer Informationen anhand von korrelationsanalytischen Verfahren

The main focus is put on developing a fuzzy data analysis method using correlation analysis and on applications in different aspects as for example strategic business planning und hospital management. After a basic overview of data analysis methods suitable for analysing fuzzy and non-fuzzy data there is given a presentation of the most relevant multivariate data analysis methods, combined with Watada/Tanaka's Fuzzy Quantification Theory (1987). Various multivariate methods such as canonicals correlation analysis, regression analysis and multidimensional scaling are being discussed. After presenting data analysis methods in the narrow sense, such as knowledge based methods, neural nets and pattern analysis it is demonstrated in what way statistical models, e.g. canonical correlation analysis, can be used as pre-analytic methods for the above mentioned data analysis methods in the narrow sense. Different multivariate methods are being compared with Fuzzy Quantification Theory proposed by Watada/Tanaka. This comparison leads to the conclusion that Fuzzy Quantification Theory is a modified version of standard multivariate methods, not suitable for analysing fuzzy data. Using Fuzzy Quantification Theory III which methodically is based on canonical correlation analysis there is proposed a new data analysis method suitable for recognizing different patterns within fuzzy data. The fuzzy information analysed is modelled by triangular fuzzy-numbers which are being integrated into the canonical correlation analysis model by a-level-cuts. Calculations which lead to the structure within the data to be analysed requires the integration of the concept of the generalized inverse as well as that of composition methods. For estimation of results the Cauchy-Schwartz-equation is used. After presenting the calculation process leading to the results of the model there is given further insight into interpreting the results. A two-stage procedure is being proposed which can reduce the quantity of the results of the model to the exact number of results necessary for pattern recognition using statistical numbers and a method developed by Röhr (1987). The relevance of the model proposed is demonstrated in the field of strategic business planning. First of all strategic business planning is discussed in the overall context of business processes. The two dimensions of a portfolio-matrix filled with artificial data are used to demonstrate the concept of the model. Using the above mentioned two-stage procedure features with identical information are recognized and eliminated. Further validation of the model is given using data from studies on work satisfaction of nurses in hospitals.

Grantor dc:publisher

Publikationsserver der RWTH Aachen University

Year dc:date

2001

Author dc:creator

• Thannheiser, Ulrike

Contributors dc:contributor

• Zimmermann, Hans-Jürgen

• info:eu-repo/classification/ddc/330
• Wirtschaft
• Datenanalyse
• Fuzzy-Menge
• Korrelationsanalyse

Statement dc:rights

• info:eu-repo/semantics/openAccess

Language dc:language

ger

Identifier

https://publications.rwth-aachen.de/search?p=id:%22RWTH-CONV-118608%22

OAI record oai:identifier

https://publications.rwth-aachen.de/oai2d?verb=GetRecord&metadataPrefix=oai_dc&identifier=oai%3Apublications.rwth-aachen.de%3A56512