Unbalanced distributed estimation and inference for the precision matrix in Gaussian graphical models

Ensiyeh Nezakati () and Eugen Pircalabelu ()
Additional contact information
Ensiyeh Nezakati: Université catholique de Louvain, LIDAM/ISBA, Belgium
Eugen Pircalabelu: Université catholique de Louvain, LIDAM/ISBA, Belgium

No 2023016, LIDAM Reprints ISBA from Université catholique de Louvain, Institute of Statistics, Biostatistics and Actuarial Sciences (ISBA)

Abstract: This paper studies the estimation of Gaussian graphical models in the unbalanced distributed framework. It provides an effective approach when the available machines are of different powers or when the existing dataset comes from different sources with different sizes and cannot be aggregated in one single machine. In this paper, we propose a new aggregated estimator of the precision matrix and justify such an approach by both theoretical and practical arguments. The limit distribution and convergence rate for this estimator are provided under sparsity conditions on the true precision matrix and controlling for the number of machines. Furthermore, a procedure for performing statistical inference is proposed. On the practical side, using a simulation study and a real data example, we show that the performance of the distributed estimator is similar to that of the non-distributed estimator that uses the full data.

Keywords: Gaussian graphical models; Precision matrix; Lasso penalization; Unbalanced distributed setting; De-biased estimator; Pseudo log-likelihood (search for similar items in EconPapers)
Pages: 23
Date: 2023-06-30
Note: In: Statistics and Computing, 2023, vol. 33, # 47
References: Add references at CitEc
Citations:

There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:aiz:louvar:2023016

DOI: 10.1007/s11222-023-10211-9

Access Statistics for this paper

More papers in LIDAM Reprints ISBA from Université catholique de Louvain, Institute of Statistics, Biostatistics and Actuarial Sciences (ISBA) Voie du Roman Pays 20, 1348 Louvain-la-Neuve (Belgium). Contact information at EDIRC.
Bibliographic data for series maintained by Nadja Peiffer ().