Compositional Data Modeling through Dirichlet Innovations

The Dirichlet distribution is a well-known candidate in modeling compositional data sets. However, the presence of outliers, fails to model such sets, making other extensions necessary. In this paper, Kummer–Dirichlet and gamma are coupled, using beta-generating technique. This development results proposal distribution, which presents greater flexibility Some general properties, as probability density functions moments presented for new candidate. method maximum likelihood applied estimation parameters. usefulness demonstrated through application synthetic real where outliers present.