Changelog

All notable changes to policytree will be documented in this file.

The format is based on Keep a Changelog and this project adheres to Semantic Versioning.

[1.2.3] - 2024-06-13

Fixed

  • Change package license to less restrictive MIT license. #172

[1.2.2] - 2023-06-23

Fixed

  • Drop C++11 compiler flag in src/Makevars per latest CRAN guidelines. #168

[1.2.1] - 2022-11-20

Fixed

  • Make the print()’d leaf labels in hybrid_policy_tree consistent with policy_tree (level first). #156

[1.2.0] - 2022-03-18

Added

  • Add hybrid_policy_tree for building deeper trees by using policy_tree to look ahead search.depth (typically equal to 2) splits. #118
  • Add some prescriptive warning messages to policy_tree in case the input type (Gamma, X) is beyond what is computationally feasible with exact tree search. #116, #129
  • Add double_robust_scores for GRF’s causal_survival_forest. #126

Fixed

  • Raise error if causal forest or instrumental forest with non-binary treatment is passed to double_robust_scores. #125

[1.1.1] - 2021-07-07

Fixed

  • Compatibility release for CRAN Solaris. #104

[1.1.0] - 2021-06-24

Changed (breaking)

IMPORTANT Some of these changes might cause small differences in results compared to previous releases, even if the same random seed is used.

  • Update policytree to use GRF version 2.0.0. The “one vs all” multi_causal_forest is deprecated and we instead utilize the new GRF estimator multi_arm_causal_forest which supports multiple treatment arms natively. multi_causal_forest will continue to work until the next release, but dispatches to multi_arm_causal_forest and emits a warning. Note that this allows for a drop-in replacement in workflows that rely on calls to double_robust_scores for policy learning, but not for workflows involving point predictions (predict(forest)) as the new GRF estimator will for K treatment arms predict a K-1 contrast matrix. #67

Added

  • Add some improved documentation around runtime and the limitations of exact tree search. #95

[1.0.4] - 2021-03-10

Added

  • Add optional leaf.label argument to plot.policy_tree allowing custom treatment names. #60
  • Add optional "type = node.id" argument to predict.policy_tree returning the leaf node the test sample falls into. #78.
  • Add optional min.node.size to policy_tree specifying the smallest permissible node size. #77

Fixed

  • Fix split.step approximation for discrete covariates by redefining to skip observations instead of unique values. Note: this may cause small differences in results compared to previous releases when using the split.step approximation parameter. #73

[1.0.3] - 2020-09-13

Fixed

  • Minor modification to C++ pow call to retain CRAN compatibility. #54

[1.0.2] - 2020-09-09

Fixed

  • Speed up prediction on very large test sets. #51

[1.0.1] - 2020-07-13

Added

  • Add support for help(<package-name>) in the R package. #41

[1.0] - 2020-06-22

Added

  • Release 1.0 has been reviewed by the Journal of Open Source Software. review, paper.

Fixed

  • Fix treatment vector categories bug. #22

[0.9.2] - 2020-03-21

Fixed

  • Minor package test modification to retain CRAN compatibility.

[0.9.1] - 2020-03-20

Added

  • Export a convenience function conditional_means
  • Runtime for tree_search is clarified. #6
  • Add an optional approximation parameter to tree_search that allow for skipping points. #5

Fixed

  • Fix predict with multi_causal_forest forest when newdata is supplied. #2
  • Fix a depth 0 base case in tree_search. #4
  • Fix pruning in the recursive case. Note that this code branch was wrong, but it would never result in the wrong tree being returned. #9

[0.9.0] - 2020-01-11

First CRAN release.

[0.1.0] - 2019-12-17

First beta release.