rename resampling iterations to 'reps' in all methods (#474)

* rename resampling iterations to 'reps' * fix test bug
py-econometrics · Jun 4, 2024 · 37bfe0f · 37bfe0f
1 parent ad61e38
commit 37bfe0f
Show file tree

Hide file tree

Showing 11 changed files with 40 additions and 40 deletions.
diff --git a/.coverage b/.coverage
diff --git a/docs/quickstart.ipynb b/docs/quickstart.ipynb
@@ -757,7 +757,7 @@
    ],
    "source": [
     "fit2 = pf.feols(fml=\"Y ~ X1\", data=data, vcov={\"CRV1\": \"group_id\"})\n",
-    "fit2.wildboottest(param=\"X1\", B=999)"
+    "fit2.wildboottest(param=\"X1\", reps=999)"
    ]
   },
   {
@@ -1219,7 +1219,7 @@
     }
    ],
    "source": [
-    "pf.rwolf([fit, fit2], param=\"X1\", B=9999, seed=1234).round(3)"
+    "pf.rwolf([fit, fit2], param=\"X1\", reps=9999, seed=1234).round(3)"
    ]
   },
   {

diff --git a/docs/replicating-the-effect.ipynb b/docs/replicating-the-effect.ipynb
@@ -439,7 +439,7 @@
    ],
    "source": [
     "fit = pf.feols(fml=\"inspection_score ~ Year + Weekend\", data=df)\n",
-    "fit.wildboottest(B=999, param=\"Year\")"
+    "fit.wildboottest(reps=999, param=\"Year\")"
    ]
   },
   {

diff --git a/pyfixest/estimation/FixestMulti_.py b/pyfixest/estimation/FixestMulti_.py
@@ -588,7 +588,7 @@ def confint(self) -> pd.DataFrame:
 
     def wildboottest(
         self,
-        B: int,
+        reps: int,
         cluster: Optional[str] = None,
         param: Optional[str] = None,
         weights_type: str = "rademacher",
@@ -642,7 +642,7 @@ def wildboottest(
             fxst = self.all_fitted_models[x]
 
             boot_res = fxst.wildboottest(
-                B,
+                reps,
                 cluster,
                 param,
                 weights_type,

diff --git a/pyfixest/estimation/estimation.py b/pyfixest/estimation/estimation.py
@@ -270,7 +270,7 @@ class for multiple models specified via `fml`.
     For example, you can run a wild (cluster) bootstrap via the `wildboottest()` method:
 
     ```{python}
-    fit.wildboottest(param = "X1", B=1000)
+    fit.wildboottest(param = "X1", reps=1000)
     ```
     would run a wild bootstrap test for the coefficient of `X1` with 1000
     bootstrap repetitions.
@@ -279,7 +279,7 @@ class for multiple models specified via `fml`.
       via the `cluster` argument:
 
     ```{python}
-    fit.wildboottest(param = "X1", B=1000, cluster="group_id")
+    fit.wildboottest(param = "X1", reps=1000, cluster="group_id")
     ```
 
     The `ritest()` method can be used to conduct randomization inference:

diff --git a/pyfixest/estimation/feols_.py b/pyfixest/estimation/feols_.py
@@ -803,7 +803,7 @@ def wald_test(self, R=None, q=None, distribution="F") -> None:
 
     def wildboottest(
         self,
-        B: int,
+        reps: int,
         cluster: Optional[str] = None,
         param: Optional[str] = None,
         weights_type: Optional[str] = "rademacher",
@@ -820,7 +820,7 @@ def wildboottest(
 
         Parameters
         ----------
-        B : int
+        reps : int
             The number of bootstrap iterations to run.
         cluster : Union[str, None], optional
             The variable used for clustering. Defaults to None. If None, then
@@ -953,7 +953,7 @@ def wildboottest(
         if run_heteroskedastic:
             inference = "HC"
 
-            boot = WildboottestHC(X=_X, Y=_Y, R=R, r=r, B=B, seed=seed)
+            boot = WildboottestHC(X=_X, Y=_Y, R=R, r=r, B=reps, seed=seed)
             boot.get_adjustments(bootstrap_type=bootstrap_type)
             boot.get_uhat(impose_null=impose_null)
             boot.get_tboot(weights_type=weights_type)
@@ -971,7 +971,7 @@ def wildboottest(
                 Y=_Y,
                 cluster=cluster_array,
                 R=R,
-                B=B,
+                B=reps,
                 seed=seed,
                 parallel=parallel,
             )
@@ -1536,7 +1536,7 @@ def confint(
         exact_match: Optional[bool] = False,
         joint: bool = False,
         seed: Optional[int] = None,
-        nboot: int = 10_000,
+        reps: int = 10_000,
     ) -> pd.DataFrame:
         r"""
         Fitted model confidence intervals.
@@ -1568,7 +1568,7 @@ def confint(
             Whether to use exact match for `keep` and `drop`. Default is False.
             If True, the pattern will be matched exactly to the coefficient name
             instead of using regular expressions.
-        nboot : int, optional
+        reps : int, optional
             The number of bootstrap iterations to run for joint confidence intervals.
             Defaults to 10_000. Only used if `joint` is True.
         seed : int, optional
@@ -1590,7 +1590,7 @@ def confint(
         data = get_data()
         fit = feols("Y ~ C(f1)", data=data)
         fit.confint(alpha=0.10).head()
-        fit.confint(alpha=0.10, joint=True, nboot=9999).head()
+        fit.confint(alpha=0.10, joint=True, reps=9999).head()
         ```
         """
         if keep is None:
@@ -1629,7 +1629,7 @@ def confint(
             D_inv = 1 / self._se[joint_indices]
             V = self._vcov[np.ix_(joint_indices, joint_indices)]
             C_coefs = (D_inv * V).T * D_inv
-            crit_val = simultaneous_crit_val(C_coefs, nboot, alpha=alpha, seed=seed)
+            crit_val = simultaneous_crit_val(C_coefs, reps, alpha=alpha, seed=seed)
 
         ub = pd.Series(
             self._beta_hat[joint_indices] + crit_val * self._se[joint_indices]

diff --git a/pyfixest/estimation/multcomp.py b/pyfixest/estimation/multcomp.py
@@ -64,7 +64,7 @@ def bonferroni(models: list[Union[Feols, Fepois]], param: str) -> pd.DataFrame:
 
 
 def rwolf(
-    models: list[Union[Feols, Fepois]], param: str, B: int, seed: int
+    models: list[Union[Feols, Fepois]], param: str, reps: int, seed: int
 ) -> pd.DataFrame:
     """
     Compute Romano-Wolf adjusted p-values for multiple hypothesis testing.
@@ -81,7 +81,7 @@ def rwolf(
         Models of type `Feiv` or `Fepois` are not supported.
     param : str
         The parameter for which the p-values should be computed.
-    B : int
+    reps : int
         The number of bootstrap replications.
     seed : int
         The seed for the random number generator.
@@ -101,11 +101,11 @@ def rwolf(
 
     data = get_data().dropna()
     fit = feols("Y ~ Y2 + X1 + X2", data=data)
-    rwolf(fit.to_list(), "X1", B=9999, seed=123)
+    rwolf(fit.to_list(), "X1", reps=9999, seed=123)
 
     fit1 = feols("Y ~ X1", data=data)
     fit2 = feols("Y ~ X1 + X2", data=data)
-    rwolf_df = rwolf([fit1, fit2], "X1", B=9999, seed=123)
+    rwolf_df = rwolf([fit1, fit2], "X1", reps=9999, seed=123)
     rwolf_df
     ```
     """
@@ -125,14 +125,14 @@ def rwolf(
 
     t_stats = all_model_stats.xs("t value").values
     t_stats = np.zeros(S)
-    boot_t_stats = np.zeros((B, S))
+    boot_t_stats = np.zeros((reps, S))
 
     for i in range(S):
         model = models[i]
 
         wildboot_res_df, bootstrapped_t_stats = model.wildboottest(
             param=param,
-            B=B,
+            reps=reps,
             return_bootstrapped_t_stats=True,
             seed=seed,  # all S iterations require the same bootstrap samples, hence seed needs to be reset
         )

diff --git a/tests/test_confint.py b/tests/test_confint.py
@@ -69,6 +69,6 @@ def test_against_doubleml():
     m = feols(
         f"y ~ -1 + {'+'.join(['X_'+str(x) for x in range(n_vars)])}", df, vcov="hetero"
     )
-    pyfixest_res = m.confint(keep="X_.$", nboot=10_000, joint=True)
+    pyfixest_res = m.confint(keep="X_.$", reps=10_000, joint=True)
 
     assert np.all(np.abs(dml_res.values - pyfixest_res.values) < 1e-2)
diff --git a/tests/test_errors.py b/tests/test_errors.py
@@ -117,7 +117,7 @@ def test_iv_errors():
         feols(fml="Y ~ 1 | Z1 ~ X1 ", vcov={"CRV3": "group_id"}, data=data)
     # wild bootstrap
     with pytest.raises(NotImplementedError):
-        feols(fml="Y ~ 1 | Z1 ~ X1 ", data=data).wildboottest(param="Z1", B=999)
+        feols(fml="Y ~ 1 | Z1 ~ X1 ", data=data).wildboottest(param="Z1", reps=999)
     # multi estimation error
     with pytest.raises(MultiEstNotSupportedError):
         feols(fml="Y + Y2 ~ 1 | Z1 ~ X1 ", data=data)
@@ -159,7 +159,7 @@ def test_wls_errors():
     data.loc[0, "weights"] = np.nan
     with pytest.raises(NotImplementedError):
         feols("Y ~ X1", data=data, weights="weights").wildboottest(
-            cluster="f1", param="X1", B=999, seed=12
+            cluster="f1", param="X1", reps=999, seed=12
         )
 
     # test for ValueError when weights are not positive
@@ -175,7 +175,7 @@ def test_wls_errors():
 
     data = get_data()
     with pytest.raises(NotImplementedError):
-        feols("Y ~ X1", data=data, weights="weights", vcov="iid").wildboottest(B=999)
+        feols("Y ~ X1", data=data, weights="weights", vcov="iid").wildboottest(reps=999)
 
 
 def test_multcomp_errors():
@@ -184,14 +184,14 @@ def test_multcomp_errors():
     # param not in model
     fit1 = feols("Y + Y2 ~ X1 | f1", data=data)
     with pytest.raises(ValueError):
-        rwolf(fit1.to_list(), param="X2", B=999, seed=92)
+        rwolf(fit1.to_list(), param="X2", reps=999, seed=92)
 
 
 def test_wildboottest_errors():
     data = get_data()
     fit = feols("Y ~ X1", data=data)
     with pytest.raises(ValueError):
-        fit.wildboottest(param="X2", B=999, seed=213)
+        fit.wildboottest(param="X2", reps=999, seed=213)
 
 
 def test_summary_errors():

diff --git a/tests/test_multcomp.py b/tests/test_multcomp.py
@@ -62,7 +62,7 @@ def test_wildrwolf():
     fit2 = feols("Y2 ~ X1", data=data)
     fit3 = feols("Y3 ~ X1", data=data)
 
-    rwolf_py = rwolf([fit1, fit2, fit3], "X1", B=9999, seed=12345)
+    rwolf_py = rwolf([fit1, fit2, fit3], "X1", reps=9999, seed=12345)
 
     # R
     fit_r = fixest.feols(ro.Formula("c(Y, Y2, Y3) ~ X1"), data=data)
@@ -81,7 +81,7 @@ def test_wildrwolf():
     fit2 = feols("Y2 ~ X1 | f1 + f2", data=data)
     fit3 = feols("Y3 ~ X1 | f1 + f2", data=data)
 
-    rwolf_py = rwolf([fit1, fit2, fit3], "X1", B=9999, seed=12345)
+    rwolf_py = rwolf([fit1, fit2, fit3], "X1", reps=9999, seed=12345)
 
     # R
     fit_r = fixest.feols(ro.Formula("c(Y, Y2, Y3) ~ X1 | f1 + f2"), data=data)

diff --git a/tests/test_wildboottest.py b/tests/test_wildboottest.py
@@ -14,39 +14,39 @@ def data():
 def test_hc_equivalence(data):
     fixest = feols(fml="Y~X2", data=data, ssc=ssc(adj=False, cluster_adj=False))
     tstat = fixest.tstat()
-    boot_tstat = fixest.wildboottest(param="X2", B=999, adj=False, cluster_adj=False)[
-        "t value"
-    ]
+    boot_tstat = fixest.wildboottest(
+        param="X2", reps=999, adj=False, cluster_adj=False
+    )["t value"]
 
     np.allclose(tstat, boot_tstat)
 
     fixest = feols(fml="Y ~ X1 + X2 | f1 + f2", data=data, vcov="hetero")
     tstat = fixest.tstat()
-    boot_tstat = fixest.wildboottest(param="X1", B=999, adj=False, cluster_adj=False)[
-        "t value"
-    ]
+    boot_tstat = fixest.wildboottest(
+        param="X1", reps=999, adj=False, cluster_adj=False
+    )["t value"]
 
     np.allclose(tstat, boot_tstat)
 
 
 def test_crv1_equivalence(data):
     fixest = feols(fml="Y~X1", data=data, vcov={"CRV1": "group_id"})
     tstat = fixest.tstat()
-    boot_tstat = fixest.wildboottest(param="X1", B=999)["t value"]
+    boot_tstat = fixest.wildboottest(param="X1", reps=999)["t value"]
 
     np.allclose(tstat, boot_tstat)
 
     fixest = feols(fml="Y ~ X1 + X2 | f1 + f2", data=data)
     tstat = fixest.tstat()
-    boot_tstat = fixest.wildboottest(param="X1", B=999, adj=False, cluster_adj=False)[
-        "t value"
-    ]
+    boot_tstat = fixest.wildboottest(
+        param="X1", reps=999, adj=False, cluster_adj=False
+    )["t value"]
 
     np.allclose(tstat, boot_tstat)
 
     fixest = feols(fml="Y ~ X1 | f1", vcov="hetero", data=data)
     boot_tstat = fixest.wildboottest(
-        param="X1", cluster="f1", B=999, adj=False, cluster_adj=False
+        param="X1", cluster="f1", reps=999, adj=False, cluster_adj=False
     )["t value"]
     tstat = fixest.vcov("hetero").tstat()