NEWS

---

CRE 0.2.5 (2023-12-06)

Added

• A copy of inTrees package source code.

Removed

• The inTrees package dependency

Added

• Add (vanilla) Stability Selection (without Error Control).
• max_rules hyper parameters for max rules filtering.
• Uncertainty Quantification in estimation by bootstrapping.
• B hyper-parameter,
• subsample hyper-parameter.
• rules(implicit form) in cre() function return.
• predict() function for ITE estimation via CRE.

Changed

• Type stability_selection binary -> string ('no','vanilla','error_control').
• Unify ntrees_gbm hyper-parameter and ntrees_gbm hyper-parameter in ntrees hyper-parameter.
• In rules generation retrieve decision rules also from internal nodes, and not just from terminal nodes.
• ite_method_dis, ite_method_inf method-parameter -> ite_method.
• ps_method_dis, ps_method_inf method-parameter -> learner_ps.
• oreg_method_dis, oreg_method_inf method-parameter -> learner_y.

Removed

• max_nodes hyper-parameter.
• Remove rules generation by Generalized Boosted Regression.
• replace hyper-parameter.
• penalty_rl hyper-parameter.
• t_pvalue hyper-parameter.
• ite_pred from cre() function return.

Bug fixes

• Error saving covariates name in CRE result when using intervention_vars.

CRE 0.2.4 (2023-06-14)

Changed

• Method paper description is updated.

CRE 0.2.3 (2023-04-27)

Removed

• Bayesian Causal Forest (bcf) ITE estimator is not supported.

CRE 0.2.2 (2023-04-17)

Changed

• Fixed failing unit tests on specific operating systems.

CRE 0.2.1 (2023-03-17)

Changed

• Replace BATE with ATE in CATE Linear Decomposition.
• Update plot() function (remove ATE, old BATE, and explicit AATEs).

Added

• Code of Conduct.

Removed

• Causal Tree benchmark in functional tests.

Bug fixes

• Rank-Deficient Rule Matrix Issue (redundant rules).
• Intervention Variables Filtering (ordered filtering).

CRE 0.2.0 (2023-01-19)

Changed

• offset method-parameter -> hyper-parameter
• estimate_ite_poisson function -> estimate_ite_tpoisson
• max_dacay hyper-parameter -> t_decay.
• interpret_select_rules function -> interpret_rules.
• generate_causal_rules function -> discover_rules.
• discover_causal_rules function ->select_rules.
• offset_name method parameter -> offset.
• Hyper and method parameters are no more required arguments for cre.
• cre object: added parameters and ite estimation.

Added

• Synthetic data set with 1 or 3 rules (generate_cre_dataset).
• S-Learner (slearner) method for ITE estimation.
• T-Learner (tlearner) method for ITE estimation.
• X-Learner (xlearner) method for ITE estimation.
• Rules Selection description in summary.cre.
• verbose parameter in summary.cre.
• ite, additional cre input parameter to use personalized ite estimations.
• Default values for hyper parameters.
• Default values for method parameters.
• Simulation experiments for estimation (estimation.R).
• Simulation experiments for discovery (discovery.R).
• extract_effect_modifiers function (utility for performance evaluation).
• evaluate function for discovery evaluation.
• confounding parameter in generate_cre_dataset to set confounding type.
• ite_pred and model in CRE results.
• binary_covariates parameter in generate_cre_dataset to set covariates domain.

Removed

• include_ps_inf method-parameter.
• include_ps_dis method-parameter.
• oreg method for ITE estimation.
• ipw method for ITE estimation.
• sipw method for ITE estimation.
• ITE standard deviation estimation.
• type_decay hyper-parameter.
• Keep only linreg for CATE estimation (remove cate_method and cate_SL_library parameters).
• method_params and hyper_params additional parameters in summary.cre.
• ite standardization for Rules Generation.
• random_state parameter.
• include_offset method parameter.

Bug fixes

• Rules Generation Issue (set rules length and fix bootstrapping).

CRE 0.1.1 (2022-10-22)

Changed

• binary parameter in generate_cre_dataset -> binary_outcome .
• filter_cate hyper-parameter -> t_pvalue.
• t_anom hyper-parameter -> t_ext.
• effect_modifier hyper-parameter -> intervention_vars.
• lasso_rules_filter function -> discover_causal_rules.
• split_data function -> honest_splitting.
• prune_rules function -> ``filter_irrelevant_rules`.
• discard_correlated_rules function -> filter_correlated_rules.
• discard_anomalous_rules function -> filter_extreme_rules.

Added

• Weighted LASSO for Causal Rules Discovery (by penalty_rl hyper-parameter).

CRE 0.1.0 (2022-10-18)

Changed

• Update examples and tests for all functions.
• q hyper-parameter -> cutoff.
• pfer_val hyper-parameter -> pfer.
• select_causal_rules function -> lasso_rules_filter.
• t hyper-parameter -> t_anom.
• Separate standardization, and remove filtering from generate_rules_matrix function.
• summary.cre function to describe results.
• min_nodes hyper-parameter -> node_size (randomForest convention).
• cre returns an S3 object.

Added

• Examples and tests for all functions.
• prune_rules function to discard un-predictive rules.
• discard_anomalous_rules function to discard anomalous rules (see t_corr hyper-parameter.).
• discard_correlated_rules function to discard correlated rules (see t_anom hyper-parameter).
• effect_modifiers parameter in generate_rules function for covariates filtering.
• generate_causal_rules function.
• Helper function with SuperLearner package for propensity score estimation in estimate_ite_xyz.
• Five methods for CATE estimation (poisson, DRLearner, bart-baggr, cf-means, linreg) in estimate_cate function.
• (ps_method_dis, ps_method_inf, or_method_dis, or_method_inf, cate_SL_library) method-parameters to complement SuperLearner package.
• cate_method method-parameter to select CATE estimation method.
• filter_cate method-parameter for estimation filtering.
• p parameter (in generate_cre_dataset function) to set the number of covariates.
• replace parameter (in generate_rules function) to allow bootstrapping.
• cre.print generic function to print cre S3 object results.
• cre.summary generic functions to summarize cre S3 object Results.
• check_input function to isolate input checks.
• estimate_ite_aipw function for augmented inverse propensity weighting.
• plot.cre generic function to plot cre S3 object results.
• test-cre_functional.R to test the functionality of the package.
• stability_selection function for causal rules selection.

Removed

• estimate_ite_blp function.
• take1() function.

Bug fixes

• Undesired 'All' Decision Rule Issue.
• No Causal Rule Selected Issue.

CRE 0.0.1

Changed

• estimate_cate include two methods for estimating the CATE values.
• cre added initial checks for binary outcome and whether to include the propensity score in the ITE estimation.
• estimate_ite_xyz conduct propensity score estimation using helper function.

Added

• Example for generate_cre_dataset.
• set_logger and get_logger.
• check_input_data function.
• generate_cre_dataset function to generate synthetic data for testing the package.
• test-generate_cre_dataset function test.
• estimate_ps function to estimate the propensity score.
• estimate_ite_xbart function to generate ITE estimates using accelerated BART.
• estimate_ite_xbcf function to generate ITE estimates using accelerated BCF.
• analyze_sensitivity function to conduct sensitivity analysis for unmeasured confounding.
• cre function to perform the entire Causal Rule Ensemble method.
• estimate_cate function to generate CATE estimates from the ITE estimates and select rules.
• estimate_ite function to generate ITE estimates using the user-specified method (calls the other estimate_ite_xyz functions).
• estimate_ite_bart function to generate ITE estimates using BART.
• estimate_ite_bcf function to generate ITE estimates using Bayesian Causal Forests.
• estimate_ite_cf function to generate ITE estimates using Causal Forests.
• estimate_ite_ipw function to generate ITE estimates using IPW.
• estimate_ite_or function to generate ITE estimates using Outcome Regression.
• estimate_ite_sipw function to generate ITE estimates using SIPW.
• extract_rules function to extract a list of causal rules from randomForest and GBM models.
• generate_rules function to generate causal rule models using randomForest and GBM methods.
• generate_rules_matrix function to convert a list of causal rules into a matrix.
• select_causal_rules function to apply penalized regression to causal rules. to select only the most important ones.
• split_data function to split input data into discovery and inference subsamples.
• take1 function to create a subsample of indices.

Removed

• seed argument in generate_cre_datase function.