×

Some estimators for compositional data. (Chinese. English summary) Zbl 1002.62042

Summary: Estimators for the mean and covariance matrix of compositional data are given by means of the seemingly unrelated regressions equations model. Especially, for compositional data sampling from the Dirichlet distribution \(D(a_1,\dots, a_{d+1})\), a quadratic estimator for the parameter \(a=\sum^{d+1}_{j=1} a_j\) and cubic estimators for \(a_1,\dots, a_{d+1}\) are proposed and the asymptotic distributions are obtained.

MSC:

62H12 Estimation in multivariate analysis
62J99 Linear inference, regression
62E20 Asymptotic distribution theory in statistics
PDFBibTeX XMLCite