|
Title
|
|
The Statistical Analysis of Compositional Data
|
|
Date / Time / Room
|
Sponsor
|
Type
|
08/12/2002
8:00 AM -
4:00 PM
Room: H-Concourse F
|
ASA, Section on Statistics & the Environment*
|
Other
|
|
Organizer:
|
n/a
|
|
Chair:
|
n/a
|
|
CE Presenter
|
- University of Glasgow
- University of Glasgow
|
Description
This course gives an introduction to theoretical and practical aspects of compositional data analysis where the data are positive vectors with constant sum, consisting of components, which are proportions of some unit. Such data occur in many disciplines in environmental studies as proportions of pollutant in water samples, in geology as major oxide compositions of rocks, in psychology and sociology as time budgets, in economics as household budget patterns, in physiology as body compositions of fat, bone, protein, etc., in medicine as renal calculi compositions. Compositional problems have been notoriously mismanaged for over a century and even today the simple methodology now available for their analysis is only slowly being recognized, many workers still applying completely inappropriate 'standard' multivariate statistics designed for unconstrained vector data. The basic problem has been a lack of understanding of the sample space - the unit simplex - and its algebraic and geometric structure. Motivated by a variety of practical compositional problems this course will set out the simple but important principles of compositional data analysis and within this framework develop a wide-ranging methodology. In analyzing the compositional data sets we shall pay particular attention to interpretation. Although computer software will not be discussed in detail during the course it is hoped to provide participants with software to enable them to perform all the statistical analyses discussed in the course.
|
|
