causal-discovery

What problem does it solve?

This Skill helps you infer a causal structure (DAG/PAG) from observational data so you can form testable causal hypotheses when experiments are not available.

Core Features & Use Cases

• Causal structure learning from data: Learn conditional-independence-based and score-based graph structures to support social science causal modeling.
• Handles latent confounding when needed: Use FCI to produce a Partial Ancestral Graph (PAG) when hidden common causes are plausible.
• Practical method selection guidance: Choose among PC, FCI, GES, and time-series-oriented options based on your data type and assumptions (cross-sectional vs. time series; sufficiency vs. latent confounding).

Quick Start

Run a DAG-structure discovery workflow on your observational dataset and return a discovered graph along with assumptions, method choice rationale, and evaluation notes.