Evaluation of Redundancy Analysis and development of a Generalized Procedure with Applications

点击翻译按钮获取中文摘要

In a variety of disciplines, there is great necesssity to analyze relationships between two sets of multivariates data. Redundancy analysis (RDA), among others, has been proposed to meet such requirement. However there have been few applications of RDA to real data and so its usefulness remained unknown. This research aimed to evaluate RDA through applied studies and also to suggest some development for RDA.Regading methodological aspect, we suggested an improved procedure (QRDA) of RDA so as to deal with multivariated data at mixed measurement levels, incorporating treatment to deal with degeneracy. We performed three applied studies. First, RDA was applied to educational data in order explore relationships between high school grade point averages and the national center examination tests. Second, RDA was utilized to examine relationships between fabric physical properties of girdles and the wearing comfort. Third, we analyzed sensory inspection data of offensive odors by using QRDA.We clarified relatioships between a set of chemical properties of the odors and another set of psychological variables in terms of sensory evaluation. Through those applications, it was concluded that RDA and QRDA are useful for data analysis.