Currently, CausalInference.jl produces (lightgraph.jl based) graphs. In order to support counterfactual inference, we need a fully specified probabilistic causal model, which means that we must express prior probability distributions over the exogenous variables.
We must therefore introduce a representation that supports distributions over these variables.
Currently,
CausalInference.jl
produces (lightgraph.jl based) graphs. In order to support counterfactual inference, we need a fully specified probabilistic causal model, which means that we must express prior probability distributions over the exogenous variables.We must therefore introduce a representation that supports distributions over these variables.