The statistical analysis of compositional data
The statistical analysis of compositional data
Constructing modal mineralogy from geochemical composition: A geometric-Bayesian approach
Computers & Geosciences
| Hi-index | 0.00 |
This contribution presents a new R package, called ''compositions''. It provides tools to analyze amount or compositional data sets in four different geometries, each one associated with an R class: rplus (for amounts, or open compositions, in a real, classical geometry), aplus (for amounts in a logarithmic geometry), rcomp (for closed compositions in a real geometry) and acomp (for closed compositions in a logistic geometry, following a log-ratio approach). The package allows to compare results obtained with these four approaches, since an analogous analysis can be performed according to each geometry, with minimal and straightforward modifications of the instructions. Beside these grounding classes, the package also includes: the most-basic features such as data transformations (e.g.logarithm, or additive logistic transform), basic statistics (both the classical ones, and those developed in the log-ratio framework of compositional analysis), high-level graphics (like ternary diagram matrix and scatter-plots) and high-level analysis (e.g.principal components or cluster analysis). Results of these functions and analysis are also provided in a consistent way among the four geometries, to ease their comparison.