Compositional covariance shrinkage and regularised partial correlations

We propose an estimation procedure for covariation in wide compositional data sets. For compositions, widely-used logratio variables are interdependent due to a common reference. Logratio uncorrelated compositions are linearly independent before the unitsum constraint is imposed. We show how they ar...

ver descrição completa

Detalhes bibliográficos
Autores: Jin, Suzanne|||0000-0001-6991-8002, Notredame, Cedric|||0000-0003-1461-0988, Erb, Ionas|||0000-0002-2331-9714
Tipo de documento: artigo
Data de publicação:2023
País:España
Recursos:Universitat Autònoma de Barcelona
Repositório:Dipòsit Digital de Documents de la UAB
Idioma:inglês
OAI Identifier:oai:ddd.uab.cat:285319
Acesso em linha:https://ddd.uab.cat/record/285319
https://dx.doi.org/urn:doi:10.57645/20.8080.02.8
Access Level:Acceso aberto
Palavra-chave:Compositional covariance structure
Logratio analysis
Partial correlation
James-Stein shrinkage
Descrição
Resumo:We propose an estimation procedure for covariation in wide compositional data sets. For compositions, widely-used logratio variables are interdependent due to a common reference. Logratio uncorrelated compositions are linearly independent before the unitsum constraint is imposed. We show how they are used to construct bespoke shrinkage targets for logratio covariance matrices and test a simple procedure for partial correlation estimates on both a simulated and a single-cell gene expression data set. For the underlying counts, different zero imputations are evaluated. The partial correlation induced by the closure is derived analytically. Data and code are available from GitHub.