Remove unused type error suppressions - core_stats

Summary:
This diff was automatically generated by the Pyre per-target upgrade tool.

It removes `# pyre-fixme` or `pyrefly: ignore` comments that are no longer needed because the underlying type errors have been resolved.
Note that it will also aim to ensure type checking runs cleanly, and will add suppressions to existing type errors.

#pyreupgrade

Differential Revision: D116539709

fbshipit-source-id: f4a73ebea111fcd71e828cff67bdeac505af979b

Author: generatedunixname89002005307016

balance/balance_frame.py

@@ -145,11 +145,8 @@ class BalanceFrame:
         False
     """
 
-    # pyre-fixme[13]: Attributes are initialized in _create() / from_frame()
     _sf_sample_pre_adjust: SampleFrame
-    # pyre-fixme[13]: Attributes are initialized in _create() / from_frame()
     _sf_sample: SampleFrame
-    # pyre-fixme[13]: Attributes are initialized in _create() / from_frame()
     _sf_target: SampleFrame | None
     # pyre-fixme[13]: Attributes are initialized in _create() / from_frame()
     _adjustment_model: dict[str, Any] | None
@@ -171,21 +168,13 @@ def _sync_sampleframe_state_from_responder(self, responder: SampleFrame) -> None
         properties stay consistent with ``_sf_sample``.
         """
         if isinstance(self, SampleFrame):
-            # pyrefly: ignore [missing-attribute]
             self._df = responder._df
-            # pyrefly: ignore [missing-attribute]
             self._id_column_name = responder._id_column_name
-            # pyrefly: ignore [missing-attribute]
             self._column_roles = responder._column_roles
-            # pyrefly: ignore [missing-attribute]
             self._weight_column_name = responder._weight_column_name
-            # pyrefly: ignore [missing-attribute]
             self._weight_metadata = responder._weight_metadata
-            # pyrefly: ignore [missing-attribute]
             self._prediction_metadata = responder._prediction_metadata
-            # pyrefly: ignore [missing-attribute]
             self._outcome_model = responder._outcome_model
-            # pyrefly: ignore [missing-attribute]
             self._df_dtypes = responder._df_dtypes
 
     @staticmethod
@@ -256,7 +245,6 @@ def _df_dtypes(self) -> pd.Series | None:
 
     @_df_dtypes.setter
     def _df_dtypes(self, value: pd.Series | None) -> None:
-        # pyrefly: ignore [missing-attribute]
         self._sf_sample._df_dtypes = value
 
     @property
@@ -295,7 +283,6 @@ def _df(self, value: pd.DataFrame | None) -> None:
                 "Cannot set _df to None. A BalanceFrame must always have a "
                 "backing DataFrame."
             )
-        # pyrefly: ignore [missing-attribute]
         self._sf_sample._df = value
 
     @property
@@ -475,7 +462,6 @@ def _create(
         instance._adjustment_history = []
         instance._links = collections.defaultdict(list)
         if target is not None:
-            # pyrefly: ignore [unsupported-operation]
             instance._links["target"] = target
 
         # When the instance is also a SampleFrame (e.g., Sample inherits
@@ -1718,7 +1704,6 @@ def _predict_outcomes_on_target(self, *, populate: bool) -> pd.DataFrame:
         self._sf_target = target_copy
         # Keep the raw-SampleFrame target link in sync with the populated copy
         # (a richer BalanceFrame/Sample link is left untouched).
-        # pyrefly: ignore [not-iterable]
         target_link = self._links.get("target") if self._links else None
         if target_link is not None and not isinstance(target_link, BalanceFrame):
             # pyrefly: ignore [unsupported-operation]
@@ -4186,7 +4171,6 @@ def _filter_sf(
             df = df.loc[:, df.columns.isin(keep_set)]
 
             new_covars = [c for c in sf._column_roles["covars"] if c in keep_set]
-            # pyrefly: ignore [missing-attribute]
             sf._column_roles = dict(sf._column_roles)
             sf._column_roles["covars"] = new_covars
             if sf._column_roles["outcomes"]:
@@ -4202,7 +4186,6 @@ def _filter_sf(
                     c for c in sf._column_roles["ignored"] if c in keep_set
                 ]
 
-        # pyrefly: ignore [missing-attribute]
         sf._df = df
         return sf
 
@@ -4523,7 +4506,6 @@ def __str__(self, pkg_source: str | None = None) -> str:
 
         if self.has_target():
             common_variables = balance_util.choose_variables(
-                # pyrefly: ignore [unsupported-operation]
                 self,
                 # pyrefly: ignore [unsupported-operation]
                 self._links["target"],

balance/cli.py

@@ -1431,11 +1431,9 @@ def main(self) -> None:
 
         logger.info("Done fitting the model, writing output")
         # Remove unneeded rows and columns
-        # pyrefly: ignore [bad-argument-type]
         output_df = self.adapt_output(output_df)
 
         # Write output
-        # pyrefly: ignore [bad-argument-type]
         self.write_outputs(output_df, diagnostics_df)
 
 

balance/datasets/loading_data.py

@@ -64,7 +64,6 @@ def _create_outcome_happiness(df: pd.DataFrame, n: int) -> npt.NDArray[np.floati
         return out
 
     if version == "01":
-        # pyrefly: ignore [bad-argument-type]
         np.random.seed(2022 - 11 - 8)  # for reproducibility
         n_target = 10000
         target_df = pd.DataFrame(
@@ -86,7 +85,6 @@ def _create_outcome_happiness(df: pd.DataFrame, n: int) -> npt.NDArray[np.floati
         # We also have missing values in gender
         target_df.loc[3:900, "gender"] = np.nan
 
-        # pyrefly: ignore [bad-argument-type]
         np.random.seed(2023 - 5 - 14)  # for reproducibility
         n_sample = 1000
         sample_df = pd.DataFrame(

balance/stats_and_plots/impact_of_weights_on_outcome.py

@@ -159,7 +159,6 @@ def weights_impact_on_outcome_ss(
         ci_lower, ci_upper = mean_diff, mean_diff
     else:
         t_stat, p_value = stats.ttest_rel(yw1, yw0, nan_policy="omit")
-        # pyrefly: ignore [missing-attribute]
         t_crit = stats.t.ppf((1 + conf_level) / 2, df=n_obs - 1)
         margin = t_crit * diff_std / np.sqrt(n_obs)
         ci_lower, ci_upper = mean_diff - margin, mean_diff + margin

balance/stats_and_plots/weighted_stats.py

@@ -16,8 +16,6 @@
 from balance.stats_and_plots.weights_stats import _check_weights_are_valid
 from balance.util import model_matrix, rm_mutual_nas
 from balance.utils.input_validation import _assert_type
-
-# pyrefly: ignore [missing-module-attribute]
 from scipy.stats import norm
 from statsmodels.stats.weightstats import DescrStatsW
 

balance/weighting_methods/cbps.py

@@ -533,7 +533,6 @@ def cbps(  # noqa
     """
 
     logger.info("Starting cbps function")
-    # pyrefly: ignore [bad-argument-type]
     np.random.seed(random_seed)  # setting random seed for cases of variations in glmnet
 
     balance_util._check_weighting_methods_input(sample_df, sample_weights, "sample")

balance/weighting_methods/ipw.py

@@ -734,7 +734,6 @@ def ipw(
 
     logger.info("Starting ipw function")
     np.random.seed(
-        # pyrefly: ignore [bad-argument-type]
         random_seed
     )  # setting random seed for cases of variations in sklearn
 

tests/test_adjustment.py

@@ -72,7 +72,6 @@ def test_trim_weights(self) -> None:
         self.assertEqual(result_weights.dtype, np.float64)
 
         # Test that no trimming parameters preserves original weights
-        # pyrefly: ignore [bad-argument-type]
         np.random.seed(42)
         random_weights = np.random.uniform(0, 1, 10000)
         untrimmed_result = trim_weights(
@@ -103,7 +102,6 @@ def test_trim_weights(self) -> None:
             trim_weights(np.array([0, 1, 2]), 1, 1)
 
         # Test weight_trimming_mean_ratio functionality
-        # pyrefly: ignore [bad-argument-type]
         np.random.seed(42)
         original_weights = np.random.uniform(0, 1, 10000)
         mean_ratio_result = trim_weights(original_weights, weight_trimming_mean_ratio=1)
@@ -119,7 +117,6 @@ def test_trim_weights(self) -> None:
         )
 
         # Test weight_trimming_percentile functionality
-        # pyrefly: ignore [bad-argument-type]
         np.random.seed(42)
         test_weights = np.random.uniform(0, 1, 10000)
 

tests/test_ascii_plots.py

@@ -844,7 +844,6 @@ def test_blog_v0_20_0_comparative_hist_population_adjusted_sample(
         blog post (``website/blog/2026/04/26/balance-0-20-0.md``)."""
         from balance.sample_class import Sample
 
-        # pyrefly: ignore [bad-argument-type]
         np.random.seed(0)
         sample_df = pd.DataFrame(
             {

tests/test_balance_frame.py

@@ -3566,7 +3566,7 @@ def test_build_adjusted_frame_migrates_legacy_ipw_fit_weight_aliases(self) -> No
         self.assertNotIn("fit_sample_weights", cleaned_model)
         self.assertNotIn("fit_target_weights", cleaned_model)
 
-    @pytest.mark.requires_sklearn_1_4  # pyre-ignore[56]
+    @pytest.mark.requires_sklearn_1_4
     @unittest.skipUnless(_SKLEARN_1_4_AVAILABLE, "requires scikit-learn >= 1.4")
     def test_store_fit_matrices_use_model_matrix_false(self) -> None:
         from sklearn.ensemble import HistGradientBoostingClassifier
@@ -5246,7 +5246,7 @@ def test_predict_weights_cbps_zero_sum_weights_raises(self) -> None:
         with self.assertRaisesRegex(ValueError, "positive sample and target weight"):
             fitted.predict_weights(data=holdout_bf)
 
-    @pytest.mark.requires_sklearn_1_4  # pyre-ignore[56]
+    @pytest.mark.requires_sklearn_1_4
     @unittest.skipUnless(_SKLEARN_1_4_AVAILABLE, "requires scikit-learn >= 1.4")
     def test_blog_v0_20_0_fit_and_predict_weights_on_holdout(self) -> None:
         """Mirrors the "Reusable fit/predict workflows" snippet from the
@@ -5319,7 +5319,7 @@ def test_blog_v0_20_0_fit_and_predict_weights_on_holdout(self) -> None:
         self.assertEqual(fitted_weights, [1.689, 1.127, 1.479, 1.083, 1.539, 1.083])
         self.assertEqual(holdout_weights, [2.709, 2.582, 2.709])
 
-    @pytest.mark.requires_sklearn_1_4  # pyre-ignore[56]
+    @pytest.mark.requires_sklearn_1_4
     @unittest.skipUnless(_SKLEARN_1_4_AVAILABLE, "requires scikit-learn >= 1.4")
     def test_blog_v0_20_0_set_fitted_model_holdout_summary(self) -> None:
         """Mirrors the ``bf_holdout.set_fitted_model(fitted)`` +
@@ -5409,7 +5409,7 @@ def test_blog_v0_20_0_set_fitted_model_holdout_summary(self) -> None:
         self.assertEqual(actual_lines, expected_lines)
 
 
-@pytest.mark.requires_sklearn_1_4  # pyre-ignore[56]
+@pytest.mark.requires_sklearn_1_4
 @unittest.skipUnless(_SKLEARN_1_4_AVAILABLE, "requires scikit-learn >= 1.4")
 class TestBlogV0_20_0SimDataHoldout(BalanceTestCase):
     """End-to-end tests for the v0.20.0 blog snippets that share a single
@@ -5538,7 +5538,6 @@ def test_blog_v0_20_0_simdata_holdout_outcomes_summary(self) -> None:
     def test_blog_v0_20_0_simdata_holdout_r_indicator(self) -> None:
         """Mirrors ``bf_holdout.weights().r_indicator()`` from the
         "r_indicator for representativeness" section."""
-        # pyrefly: ignore [missing-attribute]
         value = float(self.bf_holdout.weights().r_indicator())
         self.assertAlmostEqual(value, 0.5094452588289895, places=10)
 
@@ -5624,7 +5623,6 @@ def test_blog_v0_20_0_simdata_holdout_ascii_plot_full_output(self) -> None:
     def test_blog_v0_20_0_simdata_holdout_covars_kld(self) -> None:
         """Mirrors ``bf_holdout.covars().kld()`` from the "Distribution
         distances on raw categoricals" section."""
-        # pyrefly: ignore [missing-attribute]
         kld = self.bf_holdout.covars().kld().round(6)
         self.assertEqual(list(kld.index), ["self", "unadjusted", "unadjusted - self"])
         self.assertEqual(
@@ -7778,7 +7776,7 @@ def test_transfer_drops_stale_outcomes_hat_on_holdout(self) -> None:
         # The scored copy's responder no longer carries the stale Ŷ.
         self.assertEqual(scored._sf_sample.outcomes_hat_columns, [])
 
-    @pytest.mark.requires_sklearn_1_4  # pyre-ignore[56]
+    @pytest.mark.requires_sklearn_1_4
     @unittest.skipUnless(_SKLEARN_1_4_AVAILABLE, "requires sklearn >= 1.4")
     def test_transfer_native_categorical_auto_learner(self) -> None:
         # On sklearn>=1.4 the default learner uses the native-categorical path;