Skip to content

🔍 CausalIQ Discovery

CausalIQ Discovery provides algorithms for learning causal graph structures from observational data. There is a focus on simple, stable and competitive algorithms.

Quick Links:

✅ Current Features

  • score-based HC and Tabu algorithms
  • Support for discrete and continuous data types
  • Stability enhancements so that results not dependent on arbitrary artifacts of the data (e.g. variable order)
  • Active learning whereby algorithm can dynamically request knowledge in areas of uncertainty.

Integration with Ecosystem

  • 🎯 CausalIQ Score (causaliq-score) is used by this package to determine the score of graphs.
  • 🤖 CausalIQ Workflow uses this package in structure learning workflows.

Standalone Use

Users may use this package standalone in one-off structure learning experiments or as part of their own workflows.


CausalIQ Discovery provides the statistical foundation for the entire ecosystem, implementing proven algorithms with modern software engineering practices to support reliable causal discoverye research.