Causality

Causality in pyAgrum mainly consists in the ability to build a causal model, i.e. a (observational) Bayesian network and a set of latent variables and their relation with observation variables and in the abilidy to compute using do-calculus the causal impact in such a model.

Causality is a set of pure python3 scripts based on pyAgrum’s tools.

Note

In the figure above, pyagrum.causal module can use a LaTeX special arrow (\(\hookrightarrow\)) to compactly represent an intervention. By default, it uses the classical “do” notation. You can change this behavior to anything you want by using the following configuration keys:

pyagrum.config["causal","latex_do_prefix"]="\hookrightarrow("
pyagrum.config["causal","latex_do_suffix"]=")"

Causal models and inference

• Causal Model
  • CausalModel
• Causal Formula
  • CausalFormula
• Causal Inference
  • causalImpact()
  • doCalculusWithObservation()
  • identifyingIntervention()
• Other functions for causal inference
  • backdoor_generator()
  • frontdoor_generator()

Tutorials on causality

• Notebooks on causality in pyAgrum.

• Some implemented examples from the book of Why from Judea Pearl and Dana Mackenzie.

Tools for causality

• Exceptions
  • HedgeException
  • UnidentifiableException
• Notebook’s tools for causality
  • getCausalImpact()
  • getCausalModel()
  • showCausalImpact()
  • showCausalModel()
• Abstract Syntax Tree for Do-Calculus
  • Internal node structure
  • Basic Binary Operations
  • Complex operations