The sample space of observed data is usually explicitly or implicitly assumed to be the multidimensional real space or a subset of the same. In both cases the geometry is taken to be the usual Euclidean geometry. This basic aspect plays an essential role in the analysis of data, and very specially in the case of compositional data, where it can lead to spurious results or to poor approximations. We present a summary of the present attempts to analyse compositional data in the field of microbiome data, as well as the implications of the different approaches.

AUTHORS:  Pawlowsky-Glahn and J.J. Egozcue

SPEAKER: Pawlowsky-Glahn

EVENT: Probabilistic Microbial Modeling Symposium

San Diego (USA)

June 21-22, 2018