Skip to content

nunofachada/micompr

Repository files navigation

Checks codecov CRAN_Status_Badge Downloads docs

Summary

The micompr R package implements a procedure for comparing multivariate samples associated with different groups. The procedure uses principal component analysis to convert multivariate observations into a set of linearly uncorrelated statistical measures, which are then compared using a number of statistical methods. This technique is independent of the distributional properties of samples and automatically selects features that best explain their differences, avoiding manual selection of specific points or summary statistics. The procedure is appropriate for comparing samples of time series, images, spectrometric measures or similar multivariate observations.

How to install

Install the development version from GitHub with the following command (requires the devtools package):

devtools::install_github("nunofachada/micompr")

A stable version of the package is available on CRAN and can be installed with the following instruction:

install.packages("micompr")

Documentation

All methods and functions are fully documented and can be queried using the built-in help system. After installation, to access the man pages, invoke the micompr help page as follows:

help("micompr")

Additionally, the package contains two vignettes with a number of examples.

References

Practice

  • Fachada N, Rodrigues J, Lopes VV, Martins RC, Rosa AC. (2016) micompr: An R Package for Multivariate Independent Comparison of Observations. The R Journal 8(2):405–420. https://doi.org/10.32614/RJ-2016-055

Theory

License

MIT License