micompr: An R Package for Multivariate Independent Comparison of Observations

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.