Modelling structural zeros in compositional data via a zero-censored multivariate normal model

TL;DR

This paper introduces a novel zero-censored multivariate normal model for compositional data with structural zeros, offering an alternative to traditional methods by projecting zeros onto edges rather than vertices.

Contribution

It proposes a new modeling approach that treats zeros differently in compositional data, improving analysis of data with structural zeros.

Findings

Effective modeling of zeros in compositional data

Improved fit over existing methods

Enhanced interpretability of zero values

Abstract

We present a new model for analyzing compositional data with structural zeros. Inspired by \cite{butler2008} who suggested a model in the presence of zero values in the data we propose a model that treats the zero values in a different manner. Instead of projecting every zero value towards a vertex, we project them onto their corresponding edge and fit a zero-censored multivariate model.