Learn R Programming

abn (version 3.1.12)

export_abnFit_mle_nodes: Export node information from abnFit objects fitted with MLE (non-mixed effects)

Description

Export node information from abnFit objects fitted with MLE (non-mixed effects)

Usage

export_abnFit_mle_nodes(object, ...)

Value

A named list with two components: variables and parameters. Variables is an array where each element represents a variable with its metadata. Parameters is an array where each element represents a parameter with its coefficients.

Arguments

object

An object of class abnFit fitted with method = "mle"

...

Additional arguments (currently unused)

Details

This function extracts node parameterisation information from abnFit objects that were fitted using the Maximum Likelihood Estimation (MLE) approach without mixed-effects (i.e., no group.var specified). The function processes the coefficients and standard errors stored in the abnFit object.

The coef component contains the estimated regression coefficients for each node, stored as a matrix where column names indicate the parameter names (e.g., "g2", "m11", "b1|intercept"). These represent the linear model coefficients from the generalized linear model fitted to each node given its parents in the DAG.

The Stderror component contains the corresponding standard errors for each coefficient, providing a measure of uncertainty in the parameter estimates. The structure mirrors that of the coef component.

For different distribution types:

  • Gaussian nodes: Include intercept and slope coefficients

  • Binomial/Poisson nodes: Include intercept and slope coefficients on logit/log scale

  • Multinomial nodes: Include category-specific intercepts (reference level omitted) and coefficients, following standard multinomial logistic regression conventions