Causal Learning From Predictive Modeling for Observational Data
Nandini Ramanan, Sriraam Natarajan
Abstract
We consider the problem of learning structured causal models from observational data. In this work, we use causal Bayesian networks to represent causal relationships among model variables. To this effect, we explore the use of two types of independencies-context-specific independence (CSI) and mutual independence (MI). We use CSI to identify the candidate set of causal relationships and then use MI to quantify their strengths and construct a causal model. We validate the learned models on benchmark networks and demonstrate the effectiveness when compared to some of the state-of-the-art Causal Bayesian Network Learning algorithms from observational Data.
Topics & Concepts
Causal modelObservational studyBayesian networkMachine learningComputer scienceArtificial intelligenceIndependence (probability theory)Context (archaeology)Benchmark (surveying)Bayesian probabilityConstruct (python library)Causal structureConditional independenceSet (abstract data type)Causality (physics)MathematicsStatisticsGeographyPhysicsGeodesyQuantum mechanicsArchaeologyProgramming languageBayesian Modeling and Causal InferenceData Quality and ManagementExplainable Artificial Intelligence (XAI)