No label defined (Q3704722)
From MaRDI portal
!
This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:
scientific article; zbMATH DE number 3932184
| Language | Label | Description | Also known as |
|---|---|---|---|
| default for all languages | No label defined |
||
| English | No label defined |
scientific article; zbMATH DE number 3932184 |
Statements
1985
0 references
compositional data
0 references
neutrality
0 references
complete subcompositional independence
0 references
partition independence
0 references
distributions on the positive simplex
0 references
Dirichlet class
0 references
logistic-normal class
0 references
independence hypotheses
0 references
0.8293516635894775
0 references
0.7991080284118652
0 references
0.7943788170814514
0 references
0.7898555397987366
0 references
0.7898555397987366
0 references
The paper introduces a general class \(A^ d(\beta,\Gamma)\) of distributions on the positive simplex \(S^ d\) with probability density function \(p_ A(x| \beta,\Gamma)\) defined by \[ \log p_ A(x| \beta,\Gamma)=k(\beta,\Gamma)+\sum^{d+1}_{i=1}(\beta_ i-1)\log x_ i-2^{-1}\sum_{i\neq j}\gamma_{ij}(\log x_ i-\log x_ j)^ 2 \] where the \(\gamma_{ij}\) are symmetric, \(\Gamma =(\gamma_{ij})\), \(\beta =(\beta_ 1,...,\beta_{d+1})\), \(\sum \beta_ i\geq 0\) and the quadratic form is positive definite. This contains the Dirichlet class \((D^ d(\beta))\) as a special case for \(\gamma_{ij}=0\), \(i\neq j=1,...,d+1\) and \(\beta_ i>0\), and also the logistic-normal class \((L^ d(\beta,\Gamma))\) for \(\sum^{d+1}_{i=1}\beta_ i=0.\) The motivation for introducing the \(A^ d(\beta,\Gamma)\) class was the separateness of the \(D^ d(\beta)\) and \(L^ d(\beta,\Gamma)\) classes which resulted in difficulties in investigations of independence hypotheses. The \(L^ d(\beta,\Gamma)\) contains members with or without independence properties but no members with some of the strong forms of independence, such as complete neutrality which characterizes the \(D^ d(\beta)\) class. The new class having only one additional parameter than the \(L^ d\) class allows the development of statistical analysis methods in relation to independence hypotheses.
0 references