PARCS Documentation

PA-rtially R-andomized C-ausal S-imulator is a simulation tool for causal methods. This library is designed to facilitate simulation study design and serve as a standard benchmarking tool for causal inference and discovery methods. PARCS generates simulation mechanisms based on causal DAGs and a wide range of adjustable parameters. Once the simulation setup is described via legible instructions and rules, PARCS automatically probes the space of all complying mechanisms and synthesizes data from both observational and interventional distributions.

Install PARCS via the following command:

(.venv) $ pip install pyparcs

Citation

Note

This project is under active development.

Citation information will be put here in the near future.

Getting Started

• Theoretical background
• Creating the First Graph
• Available Functions and Distributions
• Nodes and Edges
• Partial Randomization

More Features

• Interventions and Counterfactuals
• Looping over Randomized Parameters
• Temporal Graphs
• Missing Graphs

Technical Details

• Rules and Conventions
• API
• Diagrams