Changelog
Source:NEWS.md
CRE 0.2.5 (2023-12-6)
CRAN release: 2023-12-06
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 andntrees_gbm
hyper-parameter inntrees
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
.
CRE 0.2.1 (2023-3-17)
CRAN release: 2023-03-17
Changed
- Replace BATE with ATE in CATE Linear Decomposition.
- Update
plot()
function (remove ATE, old BATE, and explicit AATEs).
CRE 0.2.0 (2023-1-19)
CRAN release: 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 insummary.cre
. -
ite
, additionalcre
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 ingenerate_cre_dataset
to set confounding type. -
ite_pred
andmodel
in CRE results. -
binary_covariates
parameter ingenerate_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 (removecate_method
andcate_SL_library
parameters). -
method_params
andhyper_params
additional parameters insummary.cre
. - ite standardization for Rules Generation.
-
random_state
parameter. -
include_offset
method parameter.
CRE 0.1.1 (2022-10-18)
CRAN release: 2022-10-22
Changed
-
binary
parameter ingenerate_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
.
CRE 0.1.0 (2022-10-17)
CRAN release: 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 (seet_corr
hyper-parameter.). -
discard_correlated_rules
function to discard correlated rules (seet_anom
hyper-parameter). -
effect_modifiers
parameter ingenerate_rules
function for covariates filtering. -
generate_causal_rules
function. - Helper function with
SuperLearner
package for propensity score estimation inestimate_ite_xyz
. - Five methods for CATE estimation (
poisson
,DRLearner
,bart-baggr
,cf-means
,linreg
) inestimate_cate
function. - (
ps_method_dis
,ps_method_inf
,or_method_dis
,or_method_inf
,cate_SL_library
) method-parameters to complementSuperLearner
package. -
cate_method
method-parameter to select CATE estimation method. -
filter_cate
method-parameter for estimation filtering. -
p
parameter (ingenerate_cre_dataset
function) to set the number of covariates. -
replace
parameter (ingenerate_rules
function) to allow bootstrapping. -
cre.print
generic function to printcre
S3 object results. -
cre.summary
generic functions to summarizecre
S3 object Results. -
check_input
function to isolate input checks. -
estimate_ite_aipw
function for augmented inverse propensity weighting. -
plot.cre
generic function to plotcre
S3 object results. -
test-cre_functional.R
to test the functionality of the package. -
stability_selection
function for causal rules selection.
CRE 0.0.1 (2021-10-20)
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
andget_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 otherestimate_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.