Normal-Block is a graphical model designed for the multivariate analysis of continuous data. It clusters variables and builds on Graphical-Lasso to infer a network of statistical dependencies between clusters.
Installation
You can install the development version of normalblockr from GitHub with:
# install.packages("pak")
pak::pak("jeannetous/normalblockr")Usage
All fitting is done using the function normal_block. An object of the class normal_data must be created (using observations Y and covariates X) and used as an input to normal_block. normal_block parameters allow to choose between sparse / not sparse models, fixed or unknown blocks…
The package comes with a small artificial data set simulated under the model.
testdata <- readRDS("testdata/testdata_normal.RDS")
Y <- testdata$Y ; X <- testdata$X
C <- testdata$parameters$C ; Q <- ncol(C)
data <- normal_data$new(Y, X)
model <- normal_block(data, blocks = C)Sparse model
model <- normal_block(data, blocks = C, sparsity = TRUE)With unobserved clustering
model <- normal_block(data, blocks = 2:5, sparsity = TRUE)
model$plot_criteria()
model_BIC <- model$get_best_model("BIC")