tidycausality

tidycausality is a development-stage R package for causal machine learning, built for seamless integration with the tidymodels ecosystem. It provides a unified, extensible framework for estimating treatment effects using modern ML techniques — starting with causal forests.

---

🔍 What is tidycausality?

tidycausality is a tidy-first, modular, and extensible package that:

• Implements causal models using parsnip-style specifications.
• Supports flexible prediction of heterogeneous treatment effects (CATEs).
• Can integrate with tuning frameworks like tune, and feature engineering tools like recipes.
• Aims for transparent, customizable modeling workflows for researchers and applied data scientists.

---

📦 Models

✅ Implemented Models

Model Name	Engine	Type	Description
causal_forest	grf	Regression	Estimates CATEs using generalized random forests (GRF).

🧪 Planned Models

Model Name	Engine	Type	Notes
bart_causal	dbarts	Regression	Bayesian Additive Regression Trees for treatment effect estimation.
x_learner	internal	Meta-learner	Decomposes effect estimation into separate models for treatment/control.
t_learner	internal	Meta-learner	Separate models for treatment and control groups.
s_learner	internal	Meta-learner	Single model with treatment as covariate.
dr_learner	internal	Doubly robust	Combines outcome modeling and propensity score modeling.
instrumental_forest	grf	IV Regression	For estimating local average treatment effects with instruments.
uplift_tree	TBD	Classification	Uplift modeling for binary outcomes and marketing campaigns.
causal_boosting	TBD	Regression	Boosted versions of causal models.
meta_stack	stacks	Ensemble	Stacking multiple causal models.

🚀 Installation

# Development version from GitHub
# install.packages("devtools")
devtools::install_github("mlatinov/tidycausality")