Logo image Logo image
  • Getting Started
  • User Guide
  • Examples
  • API reference
  • Contributing
  • Release notes
Releases
v0.7.1
v0.7
v0.6
v0.5.1
v0.5
v0.4
v0.2
v0.1.1-alpha
Branches
main
  • Foreword
  • Introduction to DoWhy
  • Modeling Causal Relations
    • Modeling cause-effect relationships with causal mechanisms
    • Diagnosing a Causal Model
      • Independence Tests
      • Refute causal structure
      • Quantifying Arrow Strength
      • Refuting Invertible Model (GCM only)
    • Learning Causal Structure from Data
  • Performing Causal Tasks
    • Estimating Causal Effects
      • Effect Estimation Using specific Effect Estimators (for ACE, mediation effect, …)
      • Estimating Average Causal Effects using GCM
    • Explaining Observed Effects and Root-Cause Analysis
      • Outlier Attribution
      • Attributing Distributional Changes
      • Quantifying Intrinsic Causal Influence
      • Unit Change Attribution
      • Feature Attribution
    • Asking and Answering What-If Questions
      • Interventions
        • Simulating the Impact of Interventions
        • Soft Interventions
      • Computing Counterfactuals
  • Miscellaneous Topics
    • Customizing Causal Mechanism Assignment
    • Estimating Confidence Intervals
    • Generate samples from a GCM
  • Citing this package

Diagnosing a Causal Model

When we modeled our problem domain as a causal model, or causal graph, a natural question that comes up: Is our causal model correct?

To answer this question, there are a number of statistical methods to verify this, which we’ll cover in the following sub-sections:

  • Independence Tests
  • Refute causal structure
  • Quantifying Arrow Strength
    • How to use it
    • Understanding the method
    • Customize the distance measure
  • Refuting Invertible Model (GCM only)

previous

Modeling cause-effect relationships with causal mechanisms

next

Independence Tests

© Copyright 2022, PyWhy contributors.

Created using Sphinx 5.3.0.