Because, the next CoDa Association conference, CoDaWork 2021, has been postponed to 2022 due to the Covid pandemics, the Toulouse team has decided to organize for the first time an online only event in 2021. It will consist in five invited talks and will take place on Thursday June 17, 2021, 14h-17h  (CEST: UTC/GMT +2h). Sessions of this event will be recorded. The recordings will be available on this website.

One of the goals is to foster the interest in the study of compositional data and their applications, to offer a forum of discussion for people concerned with the statistical treatment and modeling of compositional data or other constrained data sets, and the interpretation of models or applications involving them. Compositional data are typically defined as vectors of positive components representing parts of a whole which carry relative information. Frequently, these compositional vectors have constant sum, usually 100% or 1. These conditions render most classical statistical techniques incoherent on compositions, as they were devised for random variables with the real space as sample space. Aitchison introduced the log-ratio approach to analyze compositional data back in the eighties. His solution was based on transforming the data vector with some log-ratio transformations and applying classical techniques to the scores so obtained. This became the foundation of modern compositional data analysis, nowadays based on an appropriate geometric structure for the simplex and an appropriate representation of the sample space of compositional vectors. 

Examples of application abound in different fields of sciences: chemistry, earth and environmental sciences, life sciences and medicine with genome and microbiome research, economy, social statistics and social sciences. The following themes have been selected for the five invited talks of 2021 online CoDa-Day: geosciences, social sciences, microbiome, time-use, and theory.