Implementing causal discovery with Bayesian networks in R

In this section, we’ll explore the practical application of causal discovery using R. We’ll guide you through the step-by-step process of constructing a robust causal discovery model while leveraging various R packages to streamline the implementation, including the Bayesian-network-based approach.

Using R packages

Alright, first things first! R has many libraries that are suitable for problems in causal discovery. Each one brings something unique to the table. Let’s look at some of them as we load them into our environment:

library(pcalg)
library(bnlearn)
library(causaleffect)
library(igraph)
library(graph)

The pcalg package, released in 2009, is a solid tool for implementing causal discovery algorithms, notably the PC algorithm, and estimating causal effects. It supports flexible conditional independence tests and handles mixed data types through the micd add-on. The package also addresses...