feat: v0.9.17 — weakrobust panel + SBW + MC g-formula + end-to-end sp.causal()

Closes four gaps identified in the v1.0 gap-analysis pass:

* sp.weakrobust — Stata-style unified panel bundling Anderson-Rubin,
  Moreira CLR, Kleibergen K, Kleibergen-Paap rk LM+F, Olea-Pflueger
  effective F, and LMMP tF critical values.
* sp.sbw — Zubizarreta (2015) stable balancing weights (variance /
  entropy objective, ATT/ATC/ATE, SLSQP solver, balance table, ESS).
* sp.gformula_mc — Robins (1986) Monte-Carlo parametric g-formula
  with static or dynamic treatment strategies; complements the
  existing Bang-Robins ICE.
* sp.causal() — end-to-end orchestrator now auto-runs three new
  stages: compare_estimators (design-aware multi-estimator panel),
  sensitivity_panel (E-value + Oster δ* + Rosenbaum Γ), and cate
  (X-Learner + Causal Forest heterogeneity summary). Report gains
  sections 4b/4c/4d.

All registered in sp.* via __all__ and auto-picked-up by
sp.help / sp.list_functions / sp.describe_function.

Tests: 29 new tests in tests/test_v0917_additions.py, all green.
Full suite: 2435 passed, 4 skipped, 0 failed (18 min).

Reviewer-identified fixes applied: SBWResult now wires model_info +
_citation_key into the CausalResult parent; MC g-formula no longer
triggers the logistic loop on degenerate 0/1 columns;
_extract_effect in the workflow returns NaN when the treatment
column is missing from the params (rather than silently surfacing
the intercept).

Deferred to a separate sprint: TMLE dynamic regimes + censoring,
conformal counterfactual + weighted variants, PCMCI time-series
causal discovery, partial-ID + ML bounds, Agent-MCP integration.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: brycewang-stanford

CHANGELOG.md

@@ -2,6 +2,79 @@
 
 All notable changes to StatsPAI will be documented in this file.
 
+## [0.9.17] - 2026-04-21 — Modern-weighting + MC g-formula + weakrobust panel + end-to-end workflow
+
+Surgical release targeting four of the most-requested gaps identified
+in the v1.0 gap-analysis pass: a Stata-style unified weak-IV-robust
+diagnostic panel, the Zubizarreta (2015) stable-balancing-weights
+estimator, the Robins (1986) Monte-Carlo g-formula (complementing the
+existing Bang-Robins ICE), and a truly end-to-end `sp.causal()`
+orchestrator that auto-runs multi-estimator comparison + sensitivity
+triad + CATE heterogeneity on top of the existing diagnose/recommend/
+estimate/robustness stages.
+
+### Added
+
+- `sp.weakrobust(data, y, endog, instruments, exog)` — one-call
+  diagnostic panel that bundles Anderson-Rubin (1949), Moreira (2003)
+  Conditional LR, Kleibergen (2002) K score test, Kleibergen-Paap
+  (2006) rk LM + Wald F, Olea-Pflueger (2013) effective F, and
+  Lee-McCrary-Moreira-Porter (2022) tF critical values. `WeakRobustResult`
+  exposes `.summary()`, `.to_frame()`, and dict-style lookup. This is
+  the Python analogue of Stata 19's `estat weakrobust`, unifying
+  functionality scattered across `ivmodel` (R), `linearmodels`
+  (Python), and the Stata user-written `weakiv` / `rivtest` packages.
+
+- `sp.sbw(data, treat, covariates, y=..., estimand='att')` — Stable
+  Balancing Weights (Zubizarreta 2015 JASA). Minimises variance (or
+  KL) of the weights subject to per-covariate SMD balance tolerances
+  solved via SLSQP. Supports ATT / ATC / ATE. Reports an effective
+  sample size and before/after balance table. Complements `sp.ebalance`
+  (exact balance) and `sp.cbps` (CBPS).
+
+- `sp.gformula_mc(data, treatment_cols, confounder_cols, outcome_col)`
+  — Monte-Carlo parametric g-formula (Robins 1986). Fits per-timepoint
+  conditional models for confounders (binary logit / Gaussian OLS) and
+  simulates counterfactual trajectories under user-supplied static or
+  **dynamic** (callable) treatment strategies. Non-parametric bootstrap
+  CI. Complements the existing `sp.gformula.ice` (Bang-Robins 2005 ICE).
+
+- **Enhanced `sp.causal()` workflow** — three new stages auto-run
+  after `estimate` / `robustness`:
+  - `.compare_estimators()` — design-aware multi-estimator panel:
+    CS + SA + BJS + Wooldridge for staggered DiD; 2SLS + LIML for IV;
+    OLS + EB + CBPS + SBW + DML-PLR for observational.
+  - `.sensitivity_panel()` — E-value + Oster δ* + Rosenbaum Γ in one
+    DataFrame, matching the modern "sensitivity triad" expected by
+    top-5 econ journals.
+  - `.cate()` — X-Learner and Causal Forest heterogeneity summary
+    (per-unit CATE mean, SD, q10/q50/q90).
+  - Report output gains sections 4b / 4c / 4d.
+  - Opt-out via `CausalWorkflow.run(full=False)`; `_extract_effect`
+    helper unifies `CausalResult` and `EconometricResults` extraction.
+
+### Reviewer-identified fixes (v0.9.17 internal review)
+
+- `SBWResult.__init__` now forwards `model_info` + `_citation_key` to
+  the `CausalResult` parent, wiring it into the citation registry.
+- `MCGFormulaResult._is_binary` now requires **both** 0 and 1 levels
+  present — a degenerate column (all-0 or all-1) no longer triggers
+  the logistic Newton-Raphson loop.
+- `_extract_effect` in `CausalWorkflow` now returns NaN when the
+  treatment column is missing from the fitted params, rather than
+  silently surfacing the intercept coefficient.
+- SBW docstring clarified: reported SE is conditional-on-weights;
+  users who need full parameter-uncertainty propagation should
+  bootstrap `sp.sbw` externally.
+
+### Deferred to a separate sprint
+
+The original gap analysis also flagged TMLE dynamic regimes +
+censoring, Conformal counterfactual / weighted variants, PCMCI
+time-series causal discovery, Partial-ID + ML bounds, and the
+Agent-MCP server integration. Each is substantial enough to warrant
+its own focused sprint rather than being shipped half-finished here.
+
 ## [0.9.16] - 2026-04-20 — v1.0 breadth expansion + Bayesian family polish + Rust Phase-2 CI
 
 The largest release since the v1.0 breadth pass. Maps StatsPAI onto

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "StatsPAI"
-version = "0.9.16"
+version = "0.9.17"
 description = "The Agent-Native Causal Inference & Econometrics Toolkit for Python"
 readme = "README.md"
 license = {text = "MIT"}

src/statspai/__init__.py

@@ -22,7 +22,7 @@
 >>> sp.outreg2(result, filename="results.xlsx")
 """
 
-__version__ = "0.9.16"
+__version__ = "0.9.17"
 __author__ = "Biaoyue Wang"
 __email__ = "brycew6m@stanford.edu"
 
@@ -81,6 +81,7 @@
     OptimalMatchResult, CardinalityMatchResult,
     overlap_weights, cbps,
     genmatch, GenMatchResult,
+    sbw, SBWResult,
 )
 from .dml import dml, DoubleML, DoubleMLPLR, DoubleMLIRM, DoubleMLPLIV, DoubleMLIIVM
 from .deepiv import deepiv, DeepIV
@@ -99,7 +100,7 @@
 from .output.regression_table import regtable, RegtableResult, mean_comparison, MeanComparisonResult
 from .output.paper_tables import paper_tables, PaperTables, TEMPLATES as PAPER_TABLE_TEMPLATES
 from .postestimation import margins, marginsplot, margins_at, margins_at_plot, contrast, pwcompare, test, lincom
-from .diagnostics import oster_bounds, mccrary_test, diagnose, het_test, reset_test, vif, sensemakr, rddensity, hausman_test, anderson_rubin_test, effective_f_test, tF_critical_value, evalue, evalue_from_result, diagnose_result, estat, kitagawa_test, KitagawaResult, rosenbaum_bounds, rosenbaum_gamma, RosenbaumResult
+from .diagnostics import oster_bounds, mccrary_test, diagnose, het_test, reset_test, vif, sensemakr, rddensity, hausman_test, anderson_rubin_test, effective_f_test, tF_critical_value, evalue, evalue_from_result, diagnose_result, estat, kitagawa_test, KitagawaResult, rosenbaum_bounds, rosenbaum_gamma, RosenbaumResult, weakrobust, WeakRobustResult
 from .inference import (
     wild_cluster_bootstrap, aipw, ri_test, ipw, bootstrap, BootstrapResult,
     twoway_cluster, conley, pate, PATEEstimator, fisher_exact, FisherResult,
@@ -206,12 +207,15 @@
 # === Parametric g-formula (iterative conditional expectation) ===
 from . import gformula
 from .gformula import ice as gformula_ice_fn, ICEResult
+from .gformula import gformula_mc, MCGFormulaResult
 
 # === Target Trial Emulation (JAMA 2022 framework) ===
 from . import target_trial
+from . import target_trial as tte   # short alias
 from .target_trial import (
     protocol as target_trial_protocol,
     emulate as target_trial_emulate,
+    to_paper as target_trial_report,
     clone_censor_weight,
     immortal_time_check,
     TargetTrialProtocol,
@@ -222,6 +226,37 @@
 from . import censoring
 from .censoring import ipcw, IPCWResult
 
+# === Epidemiology primitives (OR / RR / MH / standardization / BH) ===
+from . import epi
+from .epi import (
+    odds_ratio, relative_risk, risk_difference, attributable_risk,
+    incidence_rate_ratio, number_needed_to_treat, prevalence_ratio,
+    mantel_haenszel, breslow_day_test,
+    direct_standardize, indirect_standardize,
+    bradford_hill,
+)
+
+# === Longitudinal causal inference (What If Layer 4) ===
+from . import longitudinal
+from .longitudinal import (
+    analyze as longitudinal_analyze,
+    contrast as longitudinal_contrast,
+    regime, always_treat, never_treat,
+    LongitudinalResult, Regime,
+)
+
+# === Causal-question DSL (estimand-first workflow) ===
+from . import question
+from .question import (
+    causal_question, CausalQuestion,
+    IdentificationPlan, EstimationResult,
+)
+
+# === Unified sensitivity dashboard ===
+from .robustness import (
+    unified_sensitivity, SensitivityDashboard,
+)
+
 # === Canonical datasets (consolidated facade) ===
 from . import datasets
 
@@ -305,7 +340,18 @@
 # Missing Data / Imputation
 from .imputation import mice, MICEResult, mi_estimate
 # Mendelian Randomization
-from .mendelian import mendelian_randomization, MRResult, mr_egger, mr_ivw, mr_median
+from . import mendelian
+from . import mendelian as mr   # short alias
+from .mendelian import (
+    mendelian_randomization, MRResult,
+    mr_egger, mr_ivw, mr_median,
+    mr_heterogeneity, mr_pleiotropy_egger, mr_leave_one_out,
+    mr_steiger, mr_presso, mr_radial,
+    HeterogeneityResult, PleiotropyResult, LeaveOneOutResult,
+    SteigerResult, MRPressoResult, RadialResult,
+)
+# Expose recommend_estimator at top level too
+from .dag import recommend_estimator as dag_recommend_estimator
 # Multi-cutoff / Geographic RD
 from .rd import rdmc, rdms, RDMultiResult
 # 2D Boundary RD (Cattaneo, Titiunik, Yu 2025)
@@ -526,6 +572,10 @@
     # Matching extensions
     "overlap_weights",
     "cbps",
+    "sbw",
+    "SBWResult",
+    "genmatch",
+    "GenMatchResult",
     # Inference primitives
     "subcluster_wild_bootstrap",
     "wild_cluster_ci_inv",
@@ -618,6 +668,8 @@
     "anderson_rubin_test",
     "effective_f_test",
     "tF_critical_value",
+    "weakrobust",
+    "WeakRobustResult",
     "evalue",
     "evalue_from_result",
     "diagnose_result",
@@ -974,12 +1026,36 @@
     "swig", "SWIGGraph", "SCM",
     "cevae", "CEVAE", "CEVAEResult",
     "TargetTrialProtocol", "TargetTrialResult", "CloneCensorWeightResult",
-    "target_trial_protocol", "target_trial_emulate",
-    "clone_censor_weight", "immortal_time_check",
+    "target_trial_protocol", "target_trial_emulate", "target_trial_report",
+    "clone_censor_weight", "immortal_time_check", "tte",
     "TransportWeightResult", "TransportIdentificationResult",
     "transport_generalize", "transport_weights_fn", "identify_transport",
     "OPEResult",
     "gformula_ice_fn", "ICEResult",
+    "gformula_mc", "MCGFormulaResult",
+    # v0.9.17 additions (epi primitives)
+    "epi", "odds_ratio", "relative_risk", "risk_difference",
+    "attributable_risk", "incidence_rate_ratio",
+    "number_needed_to_treat", "prevalence_ratio",
+    "mantel_haenszel", "breslow_day_test",
+    "direct_standardize", "indirect_standardize", "bradford_hill",
+    # v0.9.17 additions (MR full suite)
+    "mr", "mendelian",
+    "mr_heterogeneity", "mr_pleiotropy_egger", "mr_leave_one_out",
+    "mr_steiger", "mr_presso", "mr_radial",
+    "HeterogeneityResult", "PleiotropyResult", "LeaveOneOutResult",
+    "SteigerResult", "MRPressoResult", "RadialResult",
+    # v0.9.17 additions (longitudinal unified)
+    "longitudinal", "longitudinal_analyze", "longitudinal_contrast",
+    "regime", "always_treat", "never_treat",
+    "LongitudinalResult", "Regime",
+    # v0.9.17 additions (causal-question DSL)
+    "question", "causal_question", "CausalQuestion",
+    "IdentificationPlan", "EstimationResult",
+    # v0.9.17 additions (unified sensitivity)
+    "unified_sensitivity", "SensitivityDashboard",
+    # v0.9.17 additions (DAG UX)
+    "dag_recommend_estimator",
 ]
 
 

src/statspai/core/results.py

@@ -498,6 +498,24 @@ def __repr__(self) -> str:
         n_obs = self.data_info.get('nobs', 'Unknown')
         return f"<EconometricResults: {model_type}, {n_params} parameters, {n_obs} observations>"
 
+    def sensitivity(self, **kwargs):
+        """Run the unified sensitivity dashboard on this result.
+
+        See :func:`statspai.robustness.unified_sensitivity`.
+        """
+        from ..robustness.unified_sensitivity import unified_sensitivity
+        # Expose a 1-entry "estimate" view for compatibility
+        class _View:
+            pass
+        view = _View()
+        view.estimate = float(self.params.iloc[0])
+        view.se = float(self.std_errors.iloc[0])
+        view.ci = (float(self.conf_int_lower.iloc[0]),
+                   float(self.conf_int_upper.iloc[0]))
+        view.params = self.params
+        view.std_errors = self.std_errors
+        return unified_sensitivity(view, **kwargs)
+
 
 class CausalResult:
     """
@@ -583,6 +601,20 @@ class CausalResult:
             "  publisher={Wiley}\n"
             "}"
         ),
+        'zubizarreta_2015_sbw': (
+            "@article{zubizarreta2015stable,\n"
+            "  title={Stable weights that balance covariates for "
+            "estimation with incomplete outcome data},\n"
+            "  author={Zubizarreta, Jos{\\'e} R},\n"
+            "  journal={Journal of the American Statistical "
+            "Association},\n"
+            "  volume={110},\n"
+            "  number={511},\n"
+            "  pages={910--922},\n"
+            "  year={2015},\n"
+            "  publisher={Taylor \\& Francis}\n"
+            "}"
+        ),
         'bacon_decomposition': (
             "@article{goodman2021difference,\n"
             "  title={Difference-in-differences with variation in treatment timing},\n"
@@ -1632,6 +1664,14 @@ def __repr__(self) -> str:
     def __str__(self) -> str:
         return self.summary()
 
+    def sensitivity(self, **kwargs):
+        """Run the unified sensitivity dashboard on this result.
+
+        See :func:`statspai.robustness.unified_sensitivity`.
+        """
+        from ..robustness.unified_sensitivity import unified_sensitivity
+        return unified_sensitivity(self, **kwargs)
+
 
 # ======================================================================
 # Shared Word export helper

src/statspai/diagnostics/__init__.py

@@ -11,7 +11,10 @@
 from .sensemakr import sensemakr
 from .rddensity import rddensity
 from .hausman import hausman_test
-from .weak_iv import anderson_rubin_test, effective_f_test, tF_critical_value
+from .weak_iv import (
+    anderson_rubin_test, effective_f_test, tF_critical_value,
+    weakrobust, WeakRobustResult,
+)
 from .evalue import evalue, evalue_from_result
 from .battery import diagnose_result
 from .estat import estat
@@ -31,6 +34,8 @@
     'anderson_rubin_test',
     'effective_f_test',
     'tF_critical_value',
+    'weakrobust',
+    'WeakRobustResult',
     'evalue',
     'evalue_from_result',
     'diagnose_result',