CoDaWeb: Compositional Data Analysis Web Site

Introduction

Welcome to CoDaWeb, the compositional data web site.

The object of this web site is to serve as a forum for the exchange of information and of ideas of all those interested in compositional data analysis. This is a hot topic of research due to two main factors: (1) the long time it took to find a solution to the problem of how to perform a proper statistical analysis of this type of data, i.e. to solve the “spurious correlation problem”, as it was named by Karl Pearson back in 1897, or the “closure problem” as called by Felix Chayes in the 1960’s; and (2) its broad impact in fields as diverse as Geology, Petrology, Chemistry, Medicine, Biology, Economics, or Archaeometry, among others.

TThis web site has been planned and is maintained by the Research Group on Statistical Methods on the Simplex and other Constrained Sample Spaces, whose acronym is MEASURE. The core of the group belongs to the Computer Science and Applied Mathematics Department of the University of Girona (UdG), and includes members from the Technical University of Catalonia (UPC), the University of Extremadura (UNEX), and the Catholic University San Antonio in Murcia (UCAM). The central subject that unifies the group is the statistical analysis of compositional data following and developing the seminal work initiated by John Aitchison back in the 1980’s. The strategy we use nowadays builds on the definition of a geometric structure for the simplex (the sample space of compositional data), and consists in adapting well established statistical techniques (like cluster, discriminant, or factor analysis, regression models, etc.). Our interest lies both in real-case-studies and in the mathematical foundations, ranging from measure theory or algebraic geometry to differential calculus in the simplex. We are also developing our own statistical package, CoDaPack, due to the total lack of a specific and user-friendly software intended to treat compositional data.

If you want to access the contents of this web site, please join us registering yourself.