Unit test on fit args to conf intervals

sdfordham · May 10, 2024 · 7946b09 · 7946b09
1 parent 309720a
commit 7946b09
Show file tree

Hide file tree

Showing 3 changed files with 39 additions and 41 deletions.
diff --git a/pysyncon/inference.py b/pysyncon/inference.py
@@ -130,7 +130,7 @@ def confidence_intervals(
             raise ValueError("`step_sz_div` must be greater than 0.0")
         if scm.W is None:
             raise ValueError("No weight matrix available; fit data first.")
-        
+
         gaps = scm._gaps(Z0=Z0, Z1=Z1)
         if step_sz is None:
             if len(post_periods) > 1:

diff --git a/tests/test_conformal_interence.py b/tests/test_conformal_interence.py
@@ -24,9 +24,7 @@ def setUp(self):
             index=range(1, 5),
             columns=range(1, 11),
         )
-        self.X1 = pd.Series(
-            data=self.rng.random(size=(4,)), index=range(1, 5), name=0
-        )
+        self.X1 = pd.Series(data=self.rng.random(size=(4,)), index=range(1, 5), name=0)
         self.pre_periods = list(range(1, 21))
         self.post_periods = list(range(21, 31))
         self.max_iter = 20
@@ -218,11 +216,7 @@ def test_no_weights(self):
         }
 
         conformal_inf = ConformalInference()
-        self.assertRaises(
-            ValueError,
-            conformal_inf.confidence_intervals,
-            **kwargs
-        )
+        self.assertRaises(ValueError, conformal_inf.confidence_intervals, **kwargs)
 
     def test_step_sz_options(self):
         self.scm.fit(X0=self.X0, X1=self.X1, Z0=self.Z0, Z1=self.Z1)
@@ -240,13 +234,9 @@ def test_step_sz_options(self):
         }
 
         conformal_inf = ConformalInference()
+        conformal_inf.confidence_intervals(post_periods=self.post_periods, **kwargs)
         conformal_inf.confidence_intervals(
-            post_periods=self.post_periods,
-            **kwargs
-        )
-        conformal_inf.confidence_intervals(
-            post_periods=[self.post_periods[0]],
-            **kwargs
+            post_periods=[self.post_periods[0]], **kwargs
         )
 
     def test_root_search(self):

diff --git a/tests/test_synth.py b/tests/test_synth.py
@@ -295,59 +295,46 @@ def test_confidence_intervals(self):
         dataprep = pysyncon.Dataprep(**kwargs)
         synth = pysyncon.Synth()
         synth.fit(dataprep=dataprep)
-        
+
         # Bad option
         self.assertRaises(
             ValueError,
             synth.confidence_interval,
             alpha=0.5,
             time_periods=[4],
-            method="foo"
+            method="foo",
         )
 
         # With dataprep supplied
         try:
-            synth.confidence_interval(
-                alpha=0.5,
-                time_periods=[4],
-                dataprep=dataprep
-            )
+            synth.confidence_interval(alpha=0.5, time_periods=[4], dataprep=dataprep)
         except Exception as e:
             self.fail(f"Confidence interval failed: {e}.")
-        
+
         # Too few time periods for alpha value
         self.assertRaises(
             ValueError,
             synth.confidence_interval,
             alpha=0.05,
             time_periods=[4],
-            dataprep=dataprep
+            dataprep=dataprep,
         )
 
         # Without dataprep supplied
         try:
-            synth.confidence_interval(
-                alpha=0.5,
-                time_periods=[4]
-            )
+            synth.confidence_interval(alpha=0.5, time_periods=[4])
         except Exception as e:
             self.fail(f"Confidence interval failed: {e}.")
 
         # Too few time periods for alpha value
         self.assertRaises(
-            ValueError,
-            synth.confidence_interval,
-            alpha=0.05,
-            time_periods=[4]
+            ValueError, synth.confidence_interval, alpha=0.05, time_periods=[4]
         )
 
         # Without dataprep supplied or matrices
         synth.dataprep = None
         self.assertRaises(
-            ValueError,
-            synth.confidence_interval,
-            alpha=0.5,
-            time_periods=[4]
+            ValueError, synth.confidence_interval, alpha=0.5, time_periods=[4]
         )
 
         # No pre-periods supplied
@@ -362,7 +349,7 @@ def test_confidence_intervals(self):
             X0=X0,
             X1=X1,
             Z0=Z0,
-            Z1=Z1
+            Z1=Z1,
         )
 
         # Bad alpha value
@@ -375,9 +362,32 @@ def test_confidence_intervals(self):
             X0=X0,
             X1=X1,
             Z0=Z0,
-            Z1=Z1
+            Z1=Z1,
         )
 
+        # Add fit options
+        _, n_c = X0.shape
+        custom_V = np.full(n_c, 1 / n_c)
+        optim_method = "BFGS"
+        optim_initial = "ols"
+        optim_options = {"max_iter": 1000}
+        try:
+            synth.confidence_interval(
+                alpha=0.5,
+                time_periods=[4],
+                pre_periods=[1, 2, 3],
+                X0=X0,
+                X1=X1,
+                Z0=Z0,
+                Z1=Z1,
+                custom_V=custom_V,
+                optim_method=optim_method,
+                optim_initial=optim_initial,
+                optim_options=optim_options,
+            )
+        except Exception as e:
+            self.fail(f"Confidence interval failed: {e}.")
+
         # Dataframes supplied instead of series
         X1 = X1.to_frame()
         Z1 = Z1.to_frame()
@@ -390,7 +400,5 @@ def test_confidence_intervals(self):
             X0=X0,
             X1=X1,
             Z0=Z0,
-            Z1=Z1
+            Z1=Z1,
         )
-
-