UnCaDo: Unsure Causal Discovery

Overview & Synopsis

Most algorithms to learn causal relationships from data assume that the provided data perfectly mirrors the (in)dependencies in the system under study. This allows us to recover the correct dependence skeleton and the representative of theMarkov equivalence class of Bayesian networks that models the data. This complete partially directed acyclic graph contains some directed links that represent a direct causal influence from parent to child. In this paper we relax the mentioned requirement by allowing unsure edges in the dependence skeleton. These unsure edges can then be validated and oriented or discarded by performing experiments. We present the UnCaDo (UNsure CAusal DiscOvery) algorithm which proposes a number of necessary experiments that need to be done to gain sufficient causal information to complete the graph.