AutoCause: Python framework automates causal discovery decisions
Open-source tool records every expert choice to make causal graphs auditable
AutoCause, an open-source Python framework presented in a new paper, automates expert decisions in environmental time-series causal discovery. It wraps four established causal-discovery methods from three families, adds non-causal reference models, and grades links by method-count support. On 145 datasets from DGP-Atlas, TimeGraph, and CausalRivers, the methods recovered complementary parts of the reference graphs. Majority-supported links were more precise than single-method links on synthetic benchmarks but not against river topology. AutoCause turns inconsistent expert practice into an auditable, repeatable analysis, while causal interpretation stays with the analyst.
- AutoCause wraps 4 established causal-discovery methods from 3 families and adds non-causal reference models
- Tested on 145 datasets from DGP-Atlas, TimeGraph, and CausalRivers; majority-supported links are more precise on synthetic benchmarks
- Records every expert decision, derives defaults from a causal-audit module, and supports domain-informed overrides
Why It Matters
Makes environmental causal analysis reproducible and auditable, reducing expert bias without removing human judgment.