miniLNM

library(miniLNM)
library(dplyr)
knitr::knit_hooks$set(output = miniLNM::ansi_aware_handler)
options(crayon.enabled = TRUE)

miniLNM is a lightweight package for fitting and using logistic-normal multinomial models. It wraps a simple ‘Stan’ script (see inst/stan/ folder in the source code) and defines an S4 class that makes it easy to specify, estimate, and draw samples from the fit. For example, you can use tidyselect syntax to relate multiple compositional outputs to a set of influential biological factors.

example_data <- lnm_data()
xy <- bind_cols(example_data[c("X", "y")])
fit <- lnm(starts_with("y") ~ starts_with("x"), xy)

## Warning: Pareto k diagnostic value is 12.45. Resampling is disabled. Decreasing tol_rel_obj may help if variational algorithm has terminated prematurely. Otherwise consider using sampling instead.

The print method gives a concise summary of the fitted model, which is easier to read than the full ‘Stan’ output.

fit

You can also use predict, like in ordinary linear models, and can draw posterior predictive samples using sample.

p_hat <- predict(fit)
y_star <- sample(fit, depth = 200)

Help

We welcome questions and comments about the package either through github or email.