Model selection in canonical correlation analysis (CCA) using Bayesian model averaging

• Published in: Environmetrics (London, Ont.) Vol. 15; no. 4; pp. 291 - 311
• Main Authors: Noble, Robert, Smith, Eric P., Ye, Keying
• Format: Journal Article
• Language: English
• Published: Chichester, UK John Wiley & Sons, Ltd 01.06.2004; Wiley
• Subjects: Animal, plant and microbial ecology; Bayesian methods; Biological and medical sciences; biological monitoring; canonical correlation analysis; Earth sciences; Earth, ocean, space; Engineering and environment geology. Geothermics; Exact sciences and technology; Fundamental and applied biological sciences. Psychology; General aspects. Techniques; Methods and techniques (sampling, tagging, trapping, modelling...); Pollution, environment geology; variable selection; United States; Ohio; ecology; correlation; Bayesian methods; probability; Environmental factor; canonical correlation analysis; variable selection; indicators; North America; Biological monitoring
• ISSN: 1180-4009; 1099-095X
• DOI: 10.1002/env.641

Understanding the relationship between ecological indicators and environmental stressors is difficult. One problem is that the environmental data often consist of a large number of variables. The relationship between ecological data and environmental data is investigated using canonical correlation analysis with Bayesian model averaging used to identify important variables. A set of data from Ohio, U.S.A., is studied. Activation probabilities are used to indicate the important ecological indicators and environmental variables. A plot of model coefficients adjusted by activation probabilities is used to indicate variables that are important to canonical correlation axes. Copyright © 2004 John Wiley & Sons, Ltd.