[Repo Assist] Monthly Activity 2026-04

[github-actions]

🤖 Repo Assist here — I'm an automated AI assistant for this repository.

Activity for April 2026

Suggested Actions for Maintainer

Comprehensive list of all pending actions requiring maintainer attention (excludes items already actioned and checked off).

• Review PR [Repo Assist] Fix: improve error message when EconML estimator requires effect modifiers (X=None) #1491: fix — improve error message when EconML estimator requires effect_modifiers (X=None) (closes EconML CausalForestDML Estimation error #801) — Review
• Review PR [Repo Assist] fix: correct CausalModel.do() fit call and save _data in RegressionEstimator #1490: fix — correct CausalModel.do() (null method_params, wrong fit args, missing _data); 39 estimator tests pass (closes effect_estimate different using do operator and _estimate_effect function #309) — Review
• Review PR [Repo Assist] feat(gcm): add random_seed to interventional_samples, counterfactual_samples, average_causal_effect #1489: feat — add random_seed to interventional_samples, counterfactual_samples, average_causal_effect for reproducibility (closes Interventional samples gives inconsistent results. #1307), 3 tests pass — Review
• Review PR [Repo Assist] improve: clarify bootstrap p-value display in CausalEstimate output (closes #879) #1488: improve — clarify bootstrap p-value display with bounds + significance conclusion (closes Bootstrap-based results are ambiguous #879) — Review
• Review PR [Repo Assist] fix: WeightingSampler crashes with empty adjustment set (no covariates) #1487: fix — WeightingSampler crashes when adjustment set is empty (no covariates), 2 passing tests (closes arrays used as indices must be of integer (or boolean) type #225) — Review
• Review PR [Repo Assist] fix: correct estimate_effect_naive list-indexing bug in evaluate_effect_strength #1486: fix — correct estimate_effect_naive list-indexing bug (evaluate_effect_strength=True), with passing tests — Review
• Close PR [Repo Assist] fix: handle list-form treatment_variable in estimate_effect_naive #1461: superseded by [Repo Assist] fix: correct estimate_effect_naive list-indexing bug in evaluate_effect_strength #1486 (same fix, but [Repo Assist] fix: correct estimate_effect_naive list-indexing bug in evaluate_effect_strength #1486 adds a passing regression test) — View
• Review PR [Repo Assist] fix: correct itertools.product usage for tuple instrument_nodes in build_graph #1485: fix — correct itertools.product bug for tuple instrument_nodes in build_graph(), add regression test — Review
• Review PR [Repo Assist] fix: preserve outcome name in frontdoor/mediation estimand assumptions #1484: fix — preserve outcome name in frontdoor/mediation estimand assumptions (closes Mediation Analysis key error #214) — Review
• Review PR [Repo Assist] fix: identify all parallel mediators in identify_mediation() #1483: fix — identify all parallel mediators in identify_mediation() (closes Issue with parallel mediators/multiple directed paths #1334) — Review
• Review PR fix: identify all valid mediators in NIE/NDE estimation #1418: human contributor fix — identify all valid mediators in NIE/NDE (closes Issue with parallel mediators/multiple directed paths #1334), 4 tests pass — Review
• Review PR [Repo Assist] feat: add n_jobs parallelization to DummyOutcomeRefuter #1482: feat — add n_jobs parallelization to DummyOutcomeRefuter (closes speed up refutation by parallelization #410), 19 tests pass — Review
• Review PR [Repo Assist] fix: warn when treatment/outcome columns contain NaN values (closes #827) #1481: fix — warn when treatment/outcome columns contain NaN values (closes Check input data for NAN values before the estimation is done #827), 2 tests — Review
• Review PR [Repo Assist] test: add unit-test coverage for GraphRefuter and GraphRefutation #1479: test — add 15 unit tests for GraphRefuter / GraphRefutation (zero→covered) — Review
• Review PR [Repo Assist] fix: make CausalEstimator pickle/deepcopy-safe by excluding logger #1477: fix — make CausalEstimator pickle/deepcopy-safe by excluding logger (closes #953) — Review
• Review PR [Repo Assist] fix: pass metric-specific params via metric_params= to NearestNeighbors #1475: fix — pass metric-specific params via metric_params= to NearestNeighbors in DistanceMatchingEstimator (closes #1390), 6 new tests — Review
• Close PR [Repo Assist] fix: return scalar float p-value for single-treatment linear regression #1470: superseded by merged PR [Repo Assist] fix: return scalar float p-value for single-treatment LinearRegressionEstimator #1471 (both fix scalar p-value in LinearRegressionEstimator._test_significance for issue returning array([0.]) for p value of test_significance #1019) — View
• Review PR [Repo Assist] fix: PropensityBalanceInterpreter works with arbitrary covariate names #1472: fix — PropensityBalanceInterpreter fails with arbitrary covariate names (pd.wide_to_long convention bug + hardcoded "v0" filter), 6 new tests (closes #802) — Review
• Review PR [Repo Assist] test: add initial test coverage for gcm.fitting_sampling module #1468: test — add initial test coverage for gcm.fitting_sampling module (13 tests) — Review
• Review PR [Repo Assist] fix: support multiple treatments in PlaceboTreatmentRefuter (closes #251) #1467: fix — support multiple treatments in PlaceboTreatmentRefuter (closes #251) — Review
• Review PR [Repo Assist] test: increase sample size in linear classification auto-assignment test #1466: test — increase sample size in linear classification auto-assignment test to fix flakiness — Review
• Review PR [Repo Assist] fix: correct exact_match_cols logic in DistanceMatchingEstimator (closes #814) #1465: fix — correct exact_match_cols logic in DistanceMatchingEstimator + 12 new tests (closes #814) — Review
• Review PR [Repo Assist] feat: auto-select default estimation method when method_name is omitted #1464: feat — auto-select default estimation method when method_name is omitted (closes #365) — Review
• Review PR [Repo Assist] fix: update second-stage model estimand when pre-instantiated in TwoStageRegression #1462: fix — update second-stage model estimand when pre-instantiated in TwoStageRegression (closes #1335) — Review
• Review PR [Repo Assist] fix: correct categorical groupby KeyError and pandas FutureWarning in dummy_outcome_refuter #1460: fix — correct categorical groupby KeyError and pandas FutureWarning in dummy_outcome_refuter (closes #1316) — Review
• Review PR [Repo Assist] fix: use string key for column lookup in conditional_MI (fixes KeyError with multi-char column names) #1455: fix — use string key in conditional_MI to fix KeyError with multi-char column names (closes #949) — Review
• Review PR [Repo Assist] fix: implement asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers #1432: fix — asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers (closes #336) — Review
• Review PR [Repo Assist] perf: eliminate redundant set allocations and fix per-column kernel precision #1437: perf — eliminate redundant set allocations and fix per-column kernel precision — Review
• Review PR [Repo Assist] fix: pass encoded effect modifiers to EconML when categorical variables present #1435: fix — pass encoded effect modifiers to EconML when categorical variables present (closes #820) — Review
• Review PR [Repo Assist] fix: compute proper chi-squared p-value for CMI independence test in GraphRefuter #1431: fix — compute proper chi-squared p-value for CMI independence test in GraphRefuter (closes #413) — Review
• Review PR [Repo Assist] test: add tests for RandomCommonCause refuter #1413: test — add 6 tests for RandomCommonCause refuter — Review
• Review PR [Repo Assist] fix: preserve DataFrame index in one_hot_encode to fix data_subset_refuter bug with categorical columns #1408: fix — preserve DataFrame index in one_hot_encode (closes #1372); all encoding tests pass — Review
• Check comment EconML CausalForestDML Estimation error #801: Repo Assist explained root cause (X=None when no effect_modifiers) and workaround — View
• Check comment effect_estimate different using do operator and _estimate_effect function #309: Repo Assist explained 3 bugs in CausalModel.do() and linked fix PR — View
• Check comment Problem interpreting 95.0% confidence interval in backdoor.linear_regression #326: Repo Assist explained bootstrap CI / effect modifier mismatch and linked Delta-method PR [Repo Assist] fix: implement asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers #1432 — View
• Check comment The result of Mediation analysis with DoWhy #403: Repo Assist explained mediation non-determinism and custom treatment_value usage — View
• Check comment DoSampler: Fails when you Add a Graph to the Causal Model #83: Repo Assist explained root cause (graph ignores common_causes; empty adjustment set) and workaround — View
• Check comment Algorithms for efficient adjustment sets #464: Repo Assist noted efficient adjustment sets are already implemented — suggest closing — View
• Check comment Compatability with networkx is broken #1055: Repo Assist noted Python 3.8 dropped, networkx compat resolved — suggest closing — View
• Check comment ImportError: weighting_sampler is not an existing do sampler. #71: Repo Assist noted stale 2019 Python 3.5 issue — suggest closing — View
• Check comment Bootstrap-based results are ambiguous #879: Repo Assist explained bootstrap p-value range semantics and linked new fix PR — View
• Check comment Vectorize matching and ATE computation to speed up computation #935: Repo Assist confirmed propensity matching loop already vectorised — suggest closing — View
• Check comment distance_matching estimator raises an exception when binary treatment encoded using ints #772: Repo Assist noted PR Enabled float and int as possible binary treatment encoding options in class DistanceMatchingEstimator #987 (merged) fixed the distance_matching int encoding issue — suggest closing — View
• Check comment How can i get a deterministic output/estimate? #556: Repo Assist explained how to use np.random.seed() and random_seed= param for deterministic results — View
• Check comment AttributeError: 'CausalEstimate' object has no attribute '_estimator_object' #357: Repo Assist commented on stale _estimator_object bug — verify guidance is helpful — View
• Check comment support additional parameters in AddUnobservedCommonCause.refute_estimate #385: Repo Assist confirmed AddUnobservedCommonCause params are supported via __init__ — suggest closing — View
• Check comment arrays used as indices must be of integer (or boolean) type #225: Repo Assist explained root cause (empty covariates) and linked fix PR — View
• Check comment Mediation analysis: how to choose mediators? #249: Repo Assist linked parallel-mediators fix PR [Repo Assist] fix: identify all parallel mediators in identify_mediation() #1483 and suggested mediators_names API extension — View
• Consider closing issue Algorithms for efficient adjustment sets #464: efficient adjustment set feature already implemented — View
• Consider closing issue Compatability with networkx is broken #1055: Python 3.8 dropped, networkx compat resolved — View
• Consider closing issue ImportError: weighting_sampler is not an existing do sampler. #71: stale 2019 Python 3.5 issue — View
• Consider closing issue AttributeError: 'CausalEstimate' object has no attribute '_estimator_object' #357: stale since 2022, _estimator_object issue appears resolved — View
• Consider closing issue support additional parameters in AddUnobservedCommonCause.refute_estimate #385: parameters now fully supported, maintainer confirmed in 2022 — View
• Consider closing issue distance_matching estimator raises an exception when binary treatment encoded using ints #772: fixed by merged PR Enabled float and int as possible binary treatment encoding options in class DistanceMatchingEstimator #987 — View
• Consider closing issue Vectorize matching and ATE computation to speed up computation #935: propensity matching loop already vectorised in current code — View
• Manual action needed for issue [Repo Assist] ci: remove LLVM/Clang install step (no longer needed with scipy ≥1.15) #1443: Repo Assist cannot push workflow files; human must apply LLVM-removal patch — View
• Consider closing issue regression adjusted estimator for propensity_score_weighting #407: the doubly-robust/regression-adjusted estimator request is addressed by EconML integration

Future Work for Repo Assist

• Continue triaging open issues How the Identification Works and About Two Outcomes #240, CATE: Continuous Outcome and Binary/Categorical Variables #257, Couple of questions on causal_identifier.py #264, DoWhy Logistic Regression with Stats Api  #296, How to find ATE for continuous variables weighted by treatment?  #317, Non-linear causal relationships #320, Is there possible that estimate_conditional_effects() generate output dataframe more than 1000 rows? #346, Evaluating fitted built-in estimators on new dataset #484, How to install the GCM module of DoWHY  #513, Are there alternatives to backdoor.linear_regression for multiple continuous treatment estimation? #517, [Refute] Placebo Test #526, Add short descriptions with a code example for each of the estimation methods #530, Add short descriptions including code examples for each identification method #531 and many more without Repo Assist comments
• Follow up on human PR fix: identify all valid mediators in NIE/NDE estimation #1418 vs Repo Assist [Repo Assist] fix: identify all parallel mediators in identify_mediation() #1483 (both fix parallel mediators — needs maintainer decision)

Run History

2026-04-30 13:28 UTC — Run

• 🔧 Created PR [Repo Assist] Fix: improve error message when EconML estimator requires effect modifiers (X=None) #1491: fix — improve error message when EconML estimator requires effect_modifiers (closes #801)
• 💬 Commented on EconML CausalForestDML Estimation error #801: explained root cause (X=None when no effect_modifiers) and workaround; linked fix PR

2026-04-30 01:24 UTC — Run

• 🔧 Created PR [Repo Assist] fix: correct CausalModel.do() fit call and save _data in RegressionEstimator #1490: fix — three bugs in CausalModel.do(): null method_params crash, wrong fit() positional args, missing _data in RegressionEstimator; 1 new regression test, 39 estimator tests pass (closes #309)
• 💬 Commented on effect_estimate different using do operator and _estimate_effect function #309: explained 3 root-cause bugs in CausalModel.do() for regression estimators and linked fix PR

2026-04-29 13:12 UTC — Run

• 💬 Commented on Problem interpreting 95.0% confidence interval in backdoor.linear_regression #326: explained bootstrap CI / effect modifier mismatch (interaction term causes CI not to contain ATE); linked Delta-method PR [Repo Assist] fix: implement asymptotic CI/SE via Delta method for LinearRegressionEstimator with effect modifiers #1432
• 💬 Commented on The result of Mediation analysis with DoWhy #403: explained mediation non-determinism (missing_nodes_as_confounders ordering) and how to use custom treatment_value/control_value for mediation

2026-04-29 01:19 UTC — Run

• 💬 Commented on DoSampler: Fails when you Add a Graph to the Causal Model #83: explained that common_causes is ignored when graph is provided; traced empty adjustment set as root cause of the 0 feature(s) error; pointed to PR [Repo Assist] fix: WeightingSampler crashes with empty adjustment set (no covariates) #1487 and provided fix options
• 🔧 Created PR [Repo Assist] feat(gcm): add random_seed to interventional_samples, counterfactual_samples, average_causal_effect #1489: feat — add random_seed to interventional_samples, counterfactual_samples, average_causal_effect for reproducible results (closes #1307), 3 tests pass

2026-04-28 13:18 UTC — Run

• 🏷️ Labelled Replication of DoWhy backdoor.linear_regression with multiple treatments and effect modifiers #1367 with question
• 🏷️ Labelled Compatability with networkx is broken #1055 with enhancement
• 💬 Commented on Algorithms for efficient adjustment sets #464: noted efficient adjustment sets are already implemented (efficient-adjustment, efficient-minimal-adjustment, efficient-mincost-adjustment)
• 💬 Commented on Compatability with networkx is broken #1055: noted Python 3.8 dropped, networkx compat resolved; suggest closing
• 💬 Commented on ImportError: weighting_sampler is not an existing do sampler. #71: noted stale 2019 Python 3.5 issue; suggest closing

2026-04-28 01:18 UTC — Run

• 🔧 Created PR [Repo Assist] improve: clarify bootstrap p-value display in CausalEstimate output (closes #879) #1488: improve — clarify bootstrap p-value display in CausalEstimate.__str__ (closes #879); 5 unit tests added
• 💬 Commented on Bootstrap-based results are ambiguous #879: explained p-value range/tuple semantics and linked new PR
• 💬 Commented on Vectorize matching and ATE computation to speed up computation #935: confirmed propensity matching loop already vectorised in current code

2026-04-27 13:10 UTC — Run

• 🔧 Created PR [Repo Assist] fix: WeightingSampler crashes with empty adjustment set (no covariates) #1487: fix — WeightingSampler crashes with empty adjustment set (no covariates), 2 passing tests (closes #225)
• 💬 Commented on arrays used as indices must be of integer (or boolean) type #225: explained root cause (empty covariates list passed to sklearn/statsmodels) and linked fix PR
• 💬 Commented on Mediation analysis: how to choose mediators? #249: linked parallel-mediators fix PR [Repo Assist] fix: identify all parallel mediators in identify_mediation() #1483; noted mediators_names API extension opportunity

2026-04-27 ~01:30 UTC — Run

• 🔧 Created PR [Repo Assist] fix: correct estimate_effect_naive list-indexing bug in evaluate_effect_strength #1486: fix — correct estimate_effect_naive list-indexing bug when evaluate_effect_strength=True, with passing regression test (supersedes [Repo Assist] fix: handle list-form treatment_variable in estimate_effect_naive #1461)
• 💬 Commented on distance_matching estimator raises an exception when binary treatment encoded using ints #772: noted fix already merged in PR Enabled float and int as possible binary treatment encoding options in class DistanceMatchingEstimator #987; suggested closing
• 💬 Commented on How can i get a deterministic output/estimate? #556: explained how to use np.random.seed() and random_seed= param for reproducibility

2026-04-26 13:05 UTC — Run

• 🔧 Created PR [Repo Assist] fix: correct itertools.product usage for tuple instrument_nodes in build_graph #1485: fix — correct itertools.product bug for tuple instrument_nodes in build_graph(), add regression test
• 💬 Commented on AttributeError: 'CausalEstimate' object has no attribute '_estimator_object' #357: _estimator_object attribute is only for EconML/CausalML; issue likely stale
• 💬 Commented on support additional parameters in AddUnobservedCommonCause.refute_estimate #385: confirmed AddUnobservedCommonCause params are supported via __init__; suggested closing

2026-04-26 01:47 UTC — Run

• 🔧 Created PR [Repo Assist] fix: preserve outcome name in frontdoor/mediation estimand assumptions #1484: fix — preserve outcome name in frontdoor/mediation estimand assumption text (closes #214)

2026-04-25 13:24 UTC — Run

• 🔧 Created PR [Repo Assist] fix: identify all parallel mediators in identify_mediation() #1483: fix — identify all parallel mediators in identify_mediation() (closes #1334)

2026-04-25 01:47 UTC — Run

• 🔧 Created PR [Repo Assist] feat: add n_jobs parallelization to DummyOutcomeRefuter #1482: feat — add n_jobs parallelization to DummyOutcomeRefuter (closes #410)

2026-04-24 17:00 UTC — Run

• 🔧 Created PR [Repo Assist] test: add unit-test coverage for GraphRefuter and GraphRefutation #1479: test — add 15 unit tests for GraphRefuter / GraphRefutation
• 🔧 Created issue [Repo Assist] ci: cancel duplicate CI runs on PR pushes via concurrency group #1480: CI — add concurrency group to cancel duplicate CI runs on PR pushes

2026-04-23 18:00 UTC — Run

• 🔧 Created PR [Repo Assist] fix: make CausalEstimator pickle/deepcopy-safe by excluding logger #1477: fix — make CausalEstimator pickle/deepcopy-safe (closes #953)

2026-04-22 13:07 UTC — Run

• 🔧 Created PR [Repo Assist] fix: pass metric-specific params via metric_params= to NearestNeighbors #1475: fix — DistanceMatchingEstimator metric_params routing (closes #1390)

2026-04-21 18:00 UTC — Run

• 🔧 Created PR [Repo Assist] fix: PropensityBalanceInterpreter works with arbitrary covariate names #1472: fix — PropensityBalanceInterpreter wide_to_long / hardcoded v0 bugs (closes #802)

2026-04-18 19:00 UTC — Run

• 🔧 Created PR [Repo Assist] feat: auto-select default estimation method when method_name is omitted #1464: feat — auto-select estimation method when method_name omitted (closes #365)

2026-04-17 18:00 UTC — Run

• 🔧 Created PR [Repo Assist] fix: correct exact_match_cols logic in DistanceMatchingEstimator (closes #814) #1465: fix — correct exact_match_cols logic in DistanceMatchingEstimator (closes #814)

2026-04-15 19:00 UTC — Run

• 🔧 Created PR [Repo Assist] fix: update second-stage model estimand when pre-instantiated in TwoStageRegression #1462: fix — TwoStageRegressionEstimator pre-instantiated second_stage_model KeyError (closes #1335)
• 💬 Commented on multiple issues

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@11c9a2c442e519ff2b427bf58679f5a525353f76

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@11c9a2c442e519ff2b427bf58679f5a525353f76