colvR

From MaRDI portal
ColvR




Description

Impute missing data in count tables using a Poisson lognormal model with dimension reduction, with and without zero inflation, as described in Bricout et al. (2026) <doi:10.48550/arXiv.2602.10673>. Parameters are estimated with a variational expectation-maximization algorithm. The package also provides the asymptotic variance of the estimators, prediction intervals for the imputed counts, and model selection criteria (Bayesian information criterion, BIC, and integrated completed likelihood, ICL) to choose the dimension of the latent space.






This page was built for software: colvR