formative-ds 2.2.4__tar.gz → 2.3.0__tar.gz

{formative_ds-2.2.4 → formative_ds-2.3.0}/.github/workflows/ci.yml

@@ -16,7 +16,7 @@ jobs:
         run: uv sync --dev --group docs
 
       - name: Lint
-        run: uv run ruff check --exclude notebooks . && uv run ruff format --check --exclude notebooks .
+        run: uv run ruff check . && uv run ruff format --check --exclude notebooks .
 
       - name: Run tests
         run: uv run pytest --cov=formative --cov-report=term-missing --cov-fail-under=80

{formative_ds-2.2.4 → formative_ds-2.3.0}/CLAUDE.md

@@ -9,7 +9,7 @@ uv sync --dev                          # install dependencies
 uv run pytest                          # run all tests
 uv run pytest tests/test_iv.py         # run one file
 uv run pytest tests/test_iv.py::TestIV2SLSEstimation::test_iv_recovers_true_effect  # run one test
-make lint                              # ruff check (excludes notebooks/)
+make lint                              # ruff check
 make fmt                               # ruff format
 make check                             # lint + format check + tests with coverage (CI requires ≥80%)
 make build-docs                        # sphinx with -W: warnings are errors in CI
@@ -66,7 +66,7 @@ Do not change `_RCC_SEED` or `_PLACEBO_SEED` to 42 — the generated noise colli
 
 **Heterogeneous effects (CATE)** — `OLSObservational` and `RCT` accept `effect_modifier=` (a discrete, pre-treatment column). `fit()` then returns `OLSCATEResult`/`RCTCATEResult` with `effect_by_group`, a homogeneity F-test, and `decide_by_group()`. The headline `effect` is the sample-share-weighted average of group effects (the raw treatment coefficient under interaction coding is only the reference group's effect). Shared machinery lives in `estimators/_cate.py`; DAG validation rejects modifiers that are descendants of the treatment (mediators) or not ancestors of the outcome.
 
-**Policy learning** — `RCTResult.learn_policy(df, modifiers=[...], cost=, benefit=, max_depth=1|2)` (design note: `docs/design/policy_learning.md`) does Athey–Wager doubly robust policy learning: cross-fitted AIPW scores, exhaustive shallow-tree search over discrete features, honest (out-of-fold) value reported as the advantage over the best constant policy. Returns `PolicyResult` (`rules`, `value`, `assign()`, `refute()`). Refutation checks test the honest *value*, not tree structure — in-sample search happily splits on noise within its depth budget, so structural checks would be flaky. Trees are simplified post-hoc by dropping splits that don't change the assignment (equal-value trees with redundant conditions can win the search on candidate order).
+**Policy learning** — `RCTResult.learn_policy(df, modifiers=[...], cost=, benefit=, max_depth=1|2)` (design note: `docs/design/policy_learning.md`) does Athey–Wager doubly robust policy learning: cross-fitted AIPW scores, exhaustive shallow-tree search over discrete features, honest (out-of-fold) value reported as the advantage over the best constant policy. Returns `PolicyResult` (`rules`, `value`, `assign()`, `refute()`). Refutation checks test the honest *value*, not tree structure — in-sample search happily splits on noise within its depth budget, so structural checks would be flaky. The depth-2 search starts from the best depth-1 tree, so a deeper tree must strictly beat it — otherwise equal-value trees with redundant conditions can win on candidate order.
 
 **Package layout:**
 - `formative/causal/dag.py` — `DAG`, `_Node`
@@ -98,7 +98,7 @@ Do not change `_RCC_SEED` or `_PLACEBO_SEED` to 42 — the generated noise colli
 
 **Bootstrap-heavy tests:** matching tests use `setup_class` (not `setup_method`) so `.fit()` runs once per class. Use N=1_000 — larger N widens bootstrap SE faster than NN noise shrinks, causing refutation checks to fail even on well-specified data.
 
-**Benchmarks:** `notebooks/` holds Jupyter notebooks (LaLonde, Card–Krueger) that validate estimators against published results from famous studies; `docs/causal/benchmarks.rst` renders them as a docs page. Notebooks are excluded from ruff. Runnable per-estimator example scripts live in `examples/`.
+**Examples:** runnable per-estimator example scripts live in `examples/`.
 
 **Planned estimators:** the goal is to cover all methods in the causal inference decision tree at https://www.maxpagels.com/prototypes/causal-wizard. Remaining: **Synthetic Control**.
 

{formative_ds-2.2.4 → formative_ds-2.3.0}/Makefile

@@ -8,13 +8,13 @@ test:
 	uv run pytest --cov=formative --cov-report=term-missing --cov-fail-under=80
 
 lint:
-	uv run ruff check --exclude notebooks .
+	uv run ruff check .
 
 fmt:
-	uv run ruff format --exclude notebooks .
+	uv run ruff format .
 
 check: lint
-	uv run ruff format --check --exclude notebooks .
+	uv run ruff format --check .
 	uv run pytest --cov=formative --cov-report=term-missing --cov-fail-under=80
 
 release:  # usage: make release BUMP=patch|minor|major

{formative_ds-2.2.4 → formative_ds-2.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: formative-ds
-Version: 2.2.4
+Version: 2.3.0
 Summary: A Python library for quantitative reasoning.
 Project-URL: Homepage, https://getformative.dev
 Project-URL: Documentation, https://docs.getformative.dev

{formative_ds-2.2.4 → formative_ds-2.3.0}/docs/design/policy_learning.md

@@ -2,8 +2,8 @@
 
 **Status:** implemented (see `formative/causal/estimators/policy.py`); deviations from this
 note: refutation checks test the honest value rather than tree structure (in-sample search
-splits on noise within its depth budget, so structural checks are flaky), and final trees get
-a simplification pass that drops splits which don't change the assignment.
+splits on noise within its depth budget, so structural checks are flaky), and a deeper tree
+must strictly beat the best shallower one, so equal-value trees with redundant splits never win.
 **Date:** 2026-07-11
 
 ## Motivation
@@ -115,8 +115,7 @@ print(policy.summary())
 - Narrative rendering in `_explain.py`.
 - Tests in `tests/test_policy.py` + `tests/test_policy_refutation.py`; cross-fitting
   makes fitting expensive, so use `setup_class` like the matching tests.
-- Docs page under `docs/causal/` with a worked example; consider a benchmark notebook
-  once a public dataset with a known-good policy exists.
+- Docs page under `docs/causal/` with a worked example.
 
 ## Open questions
 

{formative_ds-2.2.4 → formative_ds-2.3.0}/docs/index.rst

@@ -29,7 +29,6 @@ under uncertainty or in strategic situations. :doc:`Get started <game/introducti
    causal/estimators
    causal/refutations
    causal/decisions
-   causal/benchmarks
 
 .. toctree::
    :maxdepth: 2

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/_base.py

@@ -1,6 +1,38 @@
 from __future__ import annotations
 
 from .._assumptions import Assumption
+from .._exceptions import IdentificationError
+
+
+def _backdoor_confounders(dag, treatment: str, outcome: str) -> set[str]:
+    """
+    Confounders under the backdoor criterion: common ancestors of treatment and
+    outcome that are not descendants of treatment (mediators or post-treatment
+    variables).
+    """
+    return (dag.ancestors(treatment) & dag.ancestors(outcome)) - dag.descendants(treatment)
+
+
+def _missing_confounders_error(missing: set[str], treatment: str, outcome: str) -> IdentificationError:
+    """``IdentificationError`` for DAG-declared confounders absent from the dataframe."""
+    return IdentificationError(
+        f"\nDAG confounders not found in dataframe: {sorted(missing)}\n\n"
+        f"Your DAG declares these variables as confounders of '{treatment}' and\n"
+        f"'{outcome}', but they are absent from the dataframe and cannot be\n"
+        f"controlled for. Note: there may also be confounders not modelled in your\n"
+        f"DAG at all — formative cannot detect those.\n\n"
+        f"Consider:\n"
+        f"  - Collecting data on {sorted(missing)} and adding it to the dataframe\n"
+        f"  - IV estimation if you have a valid instrument for '{treatment}'\n"
+        f"  - DiD or RD if a natural experiment is available"
+    )
+
+
+def _reject_missing(data, columns, action: str = "fitting") -> None:
+    """Raise if any used column has NaNs: statsmodels formulas drop those rows silently, our own code doesn't."""
+    with_nan = [c for c in dict.fromkeys(columns) if data[c].isna().any()]
+    if with_nan:
+        raise ValueError(f"Columns {with_nan} contain missing values — drop or impute them before {action}.")
 
 
 class _BaseResult:

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/_cate.py

@@ -18,6 +18,7 @@ import statsmodels.formula.api as smf
 
 from .._assumptions import Assumption
 from ..dag import DAG
+from ._base import _reject_missing
 
 CATE_ASSUMPTIONS: list[Assumption] = [
     Assumption("Correct specification of effect modification (linear interaction form)", testable=False),
@@ -86,6 +87,7 @@ def _validate_modifier_data(data: pd.DataFrame, treatment: str, modifier: str) -
     """Data checks on the effect modifier column. Raises ``ValueError`` on violation."""
     if modifier not in data.columns:
         raise ValueError(f"Effect modifier column '{modifier}' not found in dataframe.")
+    _reject_missing(data, [modifier])
     levels = data[modifier].dropna().unique()
     if len(levels) < 2:
         raise ValueError(f"Effect modifier '{modifier}' must have at least 2 levels. Found: {list(levels)}")

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/did.py

@@ -5,7 +5,7 @@ import statsmodels.formula.api as smf
 
 from .._assumptions import Assumption
 from ..dag import DAG
-from ._base import _StatsmodelsResult
+from ._base import _reject_missing, _StatsmodelsResult
 
 DID_ASSUMPTIONS: list[Assumption] = [
     Assumption(
@@ -227,6 +227,7 @@ class DiD:
         ]:
             if var not in data.columns:
                 raise ValueError(f"{label} column '{var}' not found in dataframe.")
+        _reject_missing(data, [self._group, self._time, self._outcome])
 
         for label, var in [("Group", self._group), ("Time", self._time)]:
             vals = set(data[var].dropna().unique())

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/iv.py

@@ -7,7 +7,7 @@ from statsmodels.sandbox.regression.gmm import IV2SLS as _IV2SLS
 
 from .._assumptions import Assumption
 from ..dag import DAG
-from ._base import _StatsmodelsResult
+from ._base import _backdoor_confounders, _reject_missing, _StatsmodelsResult
 
 IV_ASSUMPTIONS: list[Assumption] = [
     Assumption("Relevance: the instrument strongly affects treatment", testable=True),
@@ -228,14 +228,7 @@ class IV2SLS:
         Unobserved confounders are handled by the instrument and are not flagged
         as errors — this is the primary use case for IV estimation.
         """
-        dag = self._dag
-        T, Y = self._treatment, self._outcome
-
-        treatment_ancestors = dag.ancestors(T)
-        outcome_ancestors = dag.ancestors(Y)
-        treatment_descendants = dag.descendants(T)
-
-        confounders = (treatment_ancestors & outcome_ancestors) - treatment_descendants
+        confounders = _backdoor_confounders(self._dag, self._treatment, self._outcome)
         confounders.discard(self._instrument)
 
         return {c for c in confounders if c in data_columns}
@@ -274,6 +267,7 @@ class IV2SLS:
         controls = sorted(adjustment_set)
 
         T, Y, Z = self._treatment, self._outcome, self._instrument
+        _reject_missing(data, [T, Y, Z, *controls])
 
         # Build design matrices for 2SLS:
         #   exog:       [const, T, controls]  — includes endogenous treatment

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/matching.py

@@ -6,9 +6,8 @@ import statsmodels.api as sm
 import statsmodels.formula.api as smf
 
 from .._assumptions import Assumption
-from .._exceptions import IdentificationError
 from ..dag import DAG
-from ._base import _BaseResult
+from ._base import _backdoor_confounders, _BaseResult, _missing_confounders_error, _reject_missing
 
 MATCHING_ASSUMPTIONS: list[Assumption] = [
     Assumption("Conditional independence: no unobserved confounders given matched variables", testable=False),
@@ -283,14 +282,7 @@ class PropensityScoreMatching:
             raise ValueError("Treatment and outcome must be different variables.")
 
     def _identify(self, data_columns: set[str]) -> tuple[set[str], set[str]]:
-        dag = self._dag
-        T, Y = self._treatment, self._outcome
-
-        treatment_ancestors = dag.ancestors(T)
-        outcome_ancestors = dag.ancestors(Y)
-        treatment_descendants = dag.descendants(T)
-
-        confounders = (treatment_ancestors & outcome_ancestors) - treatment_descendants
+        confounders = _backdoor_confounders(self._dag, self._treatment, self._outcome)
         observed = {c for c in confounders if c in data_columns}
         missing = {c for c in confounders if c not in data_columns}
         return observed, missing
@@ -328,18 +320,10 @@ class PropensityScoreMatching:
         adjustment_set, missing = self._identify(data_columns)
 
         if missing:
-            raise IdentificationError(
-                f"\nDAG confounders not found in dataframe: {sorted(missing)}\n\n"
-                f"Your DAG declares these variables as confounders of "
-                f"'{self._treatment}' and '{self._outcome}', but they are "
-                f"absent from the dataframe and cannot be controlled for.\n\n"
-                f"Consider:\n"
-                f"  - Collecting data on {sorted(missing)} and adding it to the dataframe\n"
-                f"  - IV estimation if you have a valid instrument for '{self._treatment}'\n"
-                f"  - DiD or RD if a natural experiment is available"
-            )
+            raise _missing_confounders_error(missing, self._treatment, self._outcome)
 
         T, Y = self._treatment, self._outcome
+        _reject_missing(data, [T, Y, *sorted(adjustment_set)])
 
         # Unadjusted estimate: naive mean difference, no matching.
         unadjusted = data.loc[data[T] == 1, Y].mean() - data.loc[data[T] == 0, Y].mean()

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/ols.py

@@ -4,9 +4,8 @@ import pandas as pd
 import statsmodels.formula.api as smf
 
 from .._assumptions import Assumption
-from .._exceptions import IdentificationError
 from ..dag import DAG
-from ._base import _StatsmodelsResult
+from ._base import _backdoor_confounders, _missing_confounders_error, _reject_missing, _StatsmodelsResult
 from ._cate import (
     CATE_ASSUMPTIONS,
     _CATEResultMixin,
@@ -282,14 +281,7 @@ class OLSObservational:
         absent from the data. There may be additional unobserved confounders
         not represented in the DAG at all.
         """
-        dag = self._dag
-        T, Y = self._treatment, self._outcome
-
-        treatment_ancestors = dag.ancestors(T)
-        outcome_ancestors = dag.ancestors(Y)
-        treatment_descendants = dag.descendants(T)
-
-        confounders = (treatment_ancestors & outcome_ancestors) - treatment_descendants
+        confounders = _backdoor_confounders(self._dag, self._treatment, self._outcome)
 
         observed_confounders = {c for c in confounders if c in data_columns}
         dag_confounders_not_in_data = {c for c in confounders if c not in data_columns}
@@ -335,18 +327,9 @@ class OLSObservational:
         adjustment_set, dag_confounders_not_in_data = self._identify(data_columns)
 
         if dag_confounders_not_in_data:
-            raise IdentificationError(
-                f"\nDAG confounders not found in dataframe: {sorted(dag_confounders_not_in_data)}\n\n"
-                f"Your DAG declares these variables as confounders of '{self._treatment}' and\n"
-                f"'{self._outcome}', but they are absent from the dataframe and cannot be\n"
-                f"controlled for. Note: there may also be confounders not modelled in your\n"
-                f"DAG at all — formative cannot detect those.\n\n"
-                f"Consider:\n"
-                f"  - Collecting data on {sorted(dag_confounders_not_in_data)} and adding it to the dataframe\n"
-                f"  - IV estimation if you have a valid instrument for '{self._treatment}'\n"
-                f"  - DiD or RD if a natural experiment is available"
-            )
+            raise _missing_confounders_error(dag_confounders_not_in_data, self._treatment, self._outcome)
 
+        _reject_missing(data, [self._treatment, self._outcome, *sorted(adjustment_set)])
         unadjusted_result = smf.ols(f"{self._outcome} ~ {self._treatment}", data=data).fit()
 
         if self._modifier is not None:

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/policy.py

@@ -23,6 +23,7 @@ import numpy as np
 import pandas as pd
 
 from .._assumptions import Assumption
+from ._base import _reject_missing
 from ._cate import _validate_modifier_dag
 
 _FOLD_SEED = 13579
@@ -101,30 +102,6 @@ def _rules_lines(tree: PolicyNode | bool) -> list[str]:
     return [f"treat if {' and '.join(conds)}" for conds in paths] + ["otherwise don't treat"]
 
 
-def _simplify_tree(tree: PolicyNode | bool, data: pd.DataFrame) -> PolicyNode | bool:
-    """
-    Drop splits that don't change the assignment on *data*.
-
-    The exhaustive search keeps the first of several equal-value trees, which
-    can carry redundant conditions (e.g. a root split whose subtree already
-    implies it). Replacing a node by one of its children is safe whenever the
-    assignment is identical over the sample — same assignment, same value.
-    """
-    if isinstance(tree, bool):
-        return tree
-    cond = data[tree.feature].to_numpy() == tree.level
-    left = _simplify_tree(tree.if_true, data.loc[cond])
-    right = _simplify_tree(tree.if_false, data.loc[~cond])
-    if isinstance(left, bool) and isinstance(right, bool) and left == right:
-        return left
-    tree = PolicyNode(tree.feature, tree.level, left, right)
-    full = _assign_tree(tree, data)
-    for child in (left, right):
-        if np.array_equal(_assign_tree(child, data), full):
-            return child
-    return tree
-
-
 def _dr_scores(
     W: np.ndarray,
     Y: np.ndarray,
@@ -188,6 +165,9 @@ def _learn_tree(
             best_val = float(vals[i])
         return best_tree, best_val
 
+    # A deeper tree must strictly beat the best shallower one, or an equal-value
+    # tree with redundant splits can win on candidate order.
+    best_tree, best_val = _learn_tree(gamma, M, candidates, max_depth - 1)
     for i, (feature, level) in enumerate(candidates):
         mask = M[i].astype(bool)
         left_tree, left_val = _learn_tree(gamma[mask], M[:, mask], candidates, max_depth - 1)
@@ -260,7 +240,6 @@ def _fit_policy(
     value_ci = (value - 1.96 * value_se, value + 1.96 * value_se)
 
     tree, _ = _learn_tree(gamma, M, candidates, max_depth)
-    tree = _simplify_tree(tree, data)
     coverage = float(_assign_tree(tree, data).mean())
 
     return _PolicyFit(tree=tree, value=value, value_se=value_se, value_ci=value_ci, coverage=coverage)
@@ -458,10 +437,7 @@ def _validate_policy_inputs(
                 f"Policy trees need discrete features — bin continuous or high-cardinality columns first."
             )
 
-    used = [treatment, outcome, *modifiers]
-    with_nan = [c for c in used if data[c].isna().any()]
-    if with_nan:
-        raise ValueError(f"Columns {with_nan} contain missing values — drop or impute them before learning a policy.")
+    _reject_missing(data, [treatment, outcome, *modifiers], action="learning a policy")
 
     w = set(data[treatment].unique())
     if w != {0, 1}:

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/rct.py

@@ -5,7 +5,7 @@ import statsmodels.formula.api as smf
 
 from .._assumptions import Assumption
 from ..dag import DAG
-from ._base import _StatsmodelsResult
+from ._base import _reject_missing, _StatsmodelsResult
 from ._cate import (
     CATE_ASSUMPTIONS,
     _CATEResultMixin,
@@ -305,6 +305,7 @@ class RCT:
         for label, var in [("Treatment", self._treatment), ("Outcome", self._outcome)]:
             if var not in data.columns:
                 raise ValueError(f"{label} column '{var}' not found in dataframe.")
+        _reject_missing(data, [self._treatment, self._outcome])
 
         if self._modifier is not None:
             _validate_modifier_data(data, self._treatment, self._modifier)

{formative_ds-2.2.4 → formative_ds-2.3.0}/formative/causal/estimators/rdd.py

@@ -5,7 +5,7 @@ import statsmodels.formula.api as smf
 
 from .._assumptions import Assumption
 from ..dag import DAG
-from ._base import _StatsmodelsResult
+from ._base import _reject_missing, _StatsmodelsResult
 
 RDD_ASSUMPTIONS: list[Assumption] = [
     Assumption(
@@ -131,6 +131,10 @@ class RDDResult(_StatsmodelsResult):
             _check_random_common_cause,
         )
 
+        # Refute on the same rows fit() used, not the whole dataframe.
+        if self._bandwidth is not None:
+            data = data.loc[(data[self._running_var] - self._cutoff).abs() <= self._bandwidth]
+
         checks = [
             _check_placebo_cutoff(
                 data,
@@ -256,6 +260,8 @@ class RDD:
             filtered = data.loc[mask]
         else:
             filtered = data
+        # Rows outside the bandwidth are never used, so their gaps don't matter.
+        _reject_missing(filtered, [self._running_var, self._outcome])
 
         # Unadjusted effect: naive mean difference above vs below cutoff
         above = filtered.loc[filtered[self._running_var] >= self._cutoff, self._outcome].mean()

{formative_ds-2.2.4 → formative_ds-2.3.0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "formative-ds"
-version = "2.2.4"
+version = "2.3.0"
 description = "A Python library for quantitative reasoning."
 readme = "README.md"
 requires-python = ">=3.10"
@@ -37,7 +37,6 @@ Repository = "https://github.com/maxpagels/formative"
 
 [dependency-groups]
 dev  = [
-    "jupyter>=1.1.1",
     "pytest>=9,<10",
     "pytest-cov",
     "ruff>=0.9",
@@ -48,7 +47,7 @@ docs = ["sphinx>=7,<10", "sphinx-rtd-theme>=2,<4", "sphinx-copybutton>=0.5,<0.6"
 packages = ["formative"]
 
 [tool.bumpversion]
-current_version = "2.2.4"
+current_version = "2.3.0"
 commit = true
 tag = true
 tag_name = "v{new_version}"

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/_static/documentation_options.js

@@ -1,5 +1,5 @@
 const DOCUMENTATION_OPTIONS = {
-    VERSION: '2.2.3',
+    VERSION: '2.2.5',
     LANGUAGE: 'en',
     COLLAPSE_INDEX: false,
     BUILDER: 'html',

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/benchmarks.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Benchmarks &mdash; formative 2.2.3 documentation</title>
+  <title>Benchmarks &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/benchmarks.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/dag.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>DAG &mdash; formative 2.2.3 documentation</title>
+  <title>DAG &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/dag.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/decisions.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Decisions &mdash; formative 2.2.3 documentation</title>
+  <title>Decisions &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/decisions.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/estimands.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Estimands &mdash; formative 2.2.3 documentation</title>
+  <title>Estimands &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/estimands.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/estimators.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Estimators &mdash; formative 2.2.3 documentation</title>
+  <title>Estimators &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/estimators.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/introduction.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Introduction &mdash; formative 2.2.3 documentation</title>
+  <title>Introduction &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/introduction.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/motivation.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Motivation &mdash; formative 2.2.3 documentation</title>
+  <title>Motivation &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/motivation.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/refutations.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Refutations &mdash; formative 2.2.3 documentation</title>
+  <title>Refutations &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/refutations.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>

{formative_ds-2.2.4 → formative_ds-2.3.0}/site/2.2/causal/wizard.html

@@ -6,7 +6,7 @@
   <meta charset="utf-8" /><meta name="viewport" content="width=device-width, initial-scale=1" />
 
   <meta name="viewport" content="width=device-width, initial-scale=1.0" />
-  <title>Choosing an estimator &mdash; formative 2.2.3 documentation</title>
+  <title>Choosing an estimator &mdash; formative 2.2.5 documentation</title>
       <link rel="stylesheet" type="text/css" href="../_static/pygments.css?v=b86133f3" />
       <link rel="stylesheet" type="text/css" href="../_static/css/theme.css?v=9edc463e" />
       <link rel="stylesheet" type="text/css" href="../_static/copybutton.css?v=76b2166b" />
@@ -16,7 +16,7 @@
     <link rel="canonical" href="https://docs.getformative.dev/causal/wizard.html" />
       <script src="../_static/jquery.js?v=5d32c60e"></script>
       <script src="../_static/_sphinx_javascript_frameworks_compat.js?v=2cd50e6c"></script>
-      <script src="../_static/documentation_options.js?v=85e56e6b"></script>
+      <script src="../_static/documentation_options.js?v=ea147abf"></script>
       <script src="../_static/doctools.js?v=fd6eb6e6"></script>
       <script src="../_static/sphinx_highlight.js?v=6ffebe34"></script>
       <script src="../_static/clipboard.min.js?v=a7894cd8"></script>