Relational PC (RPC) is an implementation of the PC machine learning algorithm designed for use with relational data. RPC goes beyond learning statistical associations to discover causal dependencies in relational data. Given a database and schema, RPC outputs a partially directed model that represents the equivalence class of statistically indistinguishable causal models. The algorithm retains the same essential strategies employed by PC for identifying causal structure, but includes several key innovations that enable learning in relational domains.
RPC is designed and implemented by the Knowledge Discovery Laboratory in the College of Information and Computer Sciences at the University of Massachusetts Amherst. See “Learning Causal Models of Relational Domains” (Maier, et al., AAAI 2010) for additional information on RPC.