FACTOR ANALYSIS AS A MATHEMATICAL TOOL IN THE GEOSCIENCES

• Hohn Michael Edward

  West Virginia Geological Survey

Abstract

Use of factor analysis in the earth sciences has shifted in application and purpose with time since early papers in the 1960s. Early applications mirrored the goals of the originators of the methods to detect and measure actual factors that worked through some process to give observed patterns in correlations among measurable parameters. Unresolved was the issue of rotational indeterminacy, and the question: why should natural systems be modeled by linear combinations? However, many researchers discovered the utility of factor analysis and related eigenvector methods as methods for data summary; certainly the clarity of a loadings or scores plot as opposed to verbal summary of correlation matrices outweighed small losses of information and distortions resulting from projecting many dimensions to a few. Almost ignored was the issue of spatial dependence between samples.<BR>Recent years have brought new methods for handling such problems: log-linear models to resolve nonlinearity in compositional data, multi-mode factor analysis to eliminate rotational indeterminancy, and multivariate geostatistics to capture spatial dependence.

Journal

• Geoinformatics

  Geoinformatics 4 (3), 357-360, 1993

  Japan Society of Geoinformatics

Keywords

• Factor analysis
• Eigenvector
• Log-linear model
• Multivariate geostatistics

Details

• CRID: 1390001204438699776
• NII Article ID: 10004564290
• NII Book ID: AN0036643X
• DOI: 10.6010/geoinformatics1990.4.3_357
• ISSN: 1347541X
• ISSN: 0388502X
• Web Site: https://www.jstage.jst.go.jp/article/geoinformatics1990/4/3/4_3_357/_pdf
• Text Lang: en
• Data Source:
  • JaLC
  • Crossref
  • CiNii Articles
• Abstract License Flag: Disallowed