policytree

Spec sheet

Purpose

PolicyTree learns simple rule based policies represented as shallow decision trees, including interpretable treatment prescriptions.Read 12 Oct 2026 · grf-labs.github.io

Intended users

The stated applications include settings that require interpretable predictions, such as medical treatment prescription.Read 12 Oct 2026 · grf-labs.github.io

Deployment

desktopRead 23 Sep 2026 · grf-labs.github.io

Free plan

YesRead 23 Sep 2026 · grf-labs.github.io

Support

The project directs usage questions to its GitHub issues page and welcomes contributions.Read 12 Oct 2026 · grf-labs.github.io

License

The package is distributed under the MIT license with an accompanying license file.Read 12 Oct 2026 · grf-labs.github.io

Build requirements

Installing from source requires a C++ compiler and the R packages Rcpp and BH; Windows source builds require Rtools.Read 12 Oct 2026 · grf-labs.github.io

Deeper trees

hybrid_policy_tree() uses a hybrid optimal and greedy approach to fit deeper trees.Read 12 Oct 2026 · grf-labs.github.io

Installation

The latest release is available through CRAN, and the development version can be installed from GitHub.Read 12 Oct 2026 · grf-labs.github.io

Integration

PolicyTree uses reward estimates from the grf R package.Read 12 Oct 2026 · grf-labs.github.io

Intended application

The package is intended for settings requiring interpretable predictions, such as medical treatment prescription.Read 12 Oct 2026 · grf-labs.github.io

Larger datasets

The project points to fastpolicytree and sparsepolicytree as modified solvers that may improve performance on larger datasets.Read 12 Oct 2026 · grf-labs.github.io

Learning method

It uses doubly robust reward estimates to find a shallow, globally optimal decision tree.Read 12 Oct 2026 · grf-labs.github.io

Optimization

policy_tree() uses exhaustive search to fit a globally optimal tree of a specified depth that maximizes the sum of rewards.Read 12 Oct 2026 · grf-labs.github.io

Plotting

Tree plotting requires the optional DiagrammeR package.Read 12 Oct 2026 · github.com

Prediction

A fitted tree can predict an action for new observations; predictions accept numeric matrices or data frames.Read 12 Oct 2026 · github.com

Reward estimates

The double_robust_scores function computes reward estimates for a subset of grf forest types.Read 12 Oct 2026 · grf-labs.github.io

Scale considerations

The project points to fastpolicytree and sparsepolicytree as modified solvers that may improve performance on larger datasets.Read 12 Oct 2026 · grf-labs.github.io

Scale limits

The documentation says exact tree search does not scale to arbitrary dataset sizes and gives runtime examples showing substantial growth with more observations.Read 12 Oct 2026 · grf-labs.github.io

Scripting support

YesRead 23 Sep 2026 · grf-labs.github.io

Treatment types

The documentation demonstrates policy learning for binary and multi-action treatments, as well as instrumental-variable examples.Read 12 Oct 2026 · grf-labs.github.io

Tree search

The policy_tree function finds a depth k tree by exhaustive search; hybrid_policy_tree combines optimal and greedy approaches for deeper trees.Read 12 Oct 2026 · grf-labs.github.io

Where Specifiction places it

• Statistical Analysis Software#93 of 1067.1