Correcting Statistical Tests (#869)

verticapy/core/vdataframe/_corr.py

@@ -1988,15 +1988,15 @@ def pacf(
                 _executeSQL(query, print_time_sql=False)
                 vdf = create_new_vdf(tmp_view_name)
                 drop(tmp_lr0_name, method="model")
-                model = vml.LinearRegression(name=tmp_lr0_name, solver="Newton")
+                model = vml.LinearRegression(name=tmp_lr0_name, solver="newton")
                 model.fit(
                     input_relation=tmp_view_name,
                     X=[f"lag_{i}_{gen_name([column])}" for i in range(1, p)],
                     y=column,
                 )
                 model.predict(vdf, name="prediction_0")
                 drop(tmp_lr1_name, method="model")
-                model = vml.LinearRegression(name=tmp_lr1_name, solver="Newton")
+                model = vml.LinearRegression(name=tmp_lr1_name, solver="newton")
                 model.fit(
                     input_relation=tmp_view_name,
                     X=[f"lag_{i}_{gen_name([column])}" for i in range(1, p)],

verticapy/machine_learning/model_selection/statistical_tests/tsa.py

@@ -221,15 +221,18 @@ def adfuller(
                 {ts_str} AS ts 
             FROM {vdf}"""
     _executeSQL(query, print_time_sql=False)
-    model = LinearRegression(name, solver="Newton", max_iter=1000)
+    model = LinearRegression(name, solver="newton", max_iter=1000)
     predictors = ["lag1"] + [f"delta{i}" for i in range(1, p + 1)]
     if with_trend:
         predictors += ["ts"]
     try:
         model.fit(relation_name, predictors, "delta")
-        coef = model.get_vertica_attributes("details")
+    except QueryError:
+        model.set_params({"solver": "bfgs"})
+        model.fit(relation_name, predictors, "delta")
     finally:
         drop(relation_name, method="view")
+    coef = model.get_vertica_attributes("details")
     model.drop()
     if regresults:
         return coef

verticapy/tests/vDataFrame/test_vDF_descriptive_statistics.py

@@ -890,7 +890,7 @@ def test_vDF_score(self, titanic_vd):
             tol=1e-4,
             C=1.0,
             max_iter=100,
-            solver="CGD",
+            solver="cgd",
             penalty="ENet",
             l1_ratio=0.5,
         )

verticapy/tests/vDataFrame/test_vDF_plot_plotly.py

@@ -313,7 +313,7 @@ def champion_challenger_plot_result(load_plotly, titanic_vd):
 @pytest.fixture(scope="class")
 def stepwise_plot_result(load_plotly, titanic_vd):
     model = LogisticRegression(
-        name="test_LR_titanic", tol=1e-4, max_iter=100, solver="Newton"
+        name="test_LR_titanic", tol=1e-4, max_iter=100, solver="newton"
     )
     stepwise_result = stepwise(
         model,
@@ -2645,7 +2645,7 @@ def test_additional_options_custom_height(self, load_plotly, titanic_vd):
         custom_height = 650
         custom_width = 700
         model = LogisticRegression(
-            name="test_LR_titanic", tol=1e-4, max_iter=100, solver="Newton"
+            name="test_LR_titanic", tol=1e-4, max_iter=100, solver="newton"
         )
         # Act
         stepwise_result = stepwise(

verticapy/tests_new/machine_learning/vertica/conftest.py

@@ -105,7 +105,7 @@ def _get_vpy_model(model_class, X=None, y=None, **kwargs):
             if model_class in ["Lasso", "ElasticNet"]:
                 solver = "cgd"
             else:
-                solver = "Newton"
+                solver = "newton"
 
         if model_class in ["RandomForestRegressor", "RandomForestClassifier"]:
             model = getattr(vpy_tree, model_class)(

verticapy/tests_new/mlops/test_model_tracking.py

@@ -146,7 +146,7 @@ def test_load_best_model(self, experiment, best_model_name, metric):
 
 @pytest.fixture(scope="module")
 def reg_model1(winequality_vpy):
-    model = LinearRegression("reg_m1", solver="BFGS", max_iter=1)
+    model = LinearRegression("reg_m1", solver="bfgs", max_iter=1)
     model.drop()
 
     model.fit(
@@ -160,7 +160,7 @@ def reg_model1(winequality_vpy):
 
 @pytest.fixture(scope="module")
 def reg_model2(winequality_vpy):
-    model = LinearRegression("reg_m2", solver="BFGS", max_iter=3)
+    model = LinearRegression("reg_m2", solver="bfgs", max_iter=3)
     model.drop()
 
     model.fit(
@@ -278,7 +278,7 @@ def test_load_best_model(self, regressor_experiment):
 
 @pytest.fixture(scope="module")
 def bin_model1(winequality_vpy):
-    model = LogisticRegression("bin_m1", solver="Newton", max_iter=5, penalty=None)
+    model = LogisticRegression("bin_m1", solver="newton", max_iter=5, penalty=None)
     model.drop()
 
     model.fit(
@@ -292,7 +292,7 @@ def bin_model1(winequality_vpy):
 
 @pytest.fixture(scope="module")
 def bin_model2(winequality_vpy):
-    model = LogisticRegression("bin_m2", solver="BFGS", max_iter=5, penalty=None)
+    model = LogisticRegression("bin_m2", solver="bfgs", max_iter=5, penalty=None)
     model.drop()
 
     model.fit(

verticapy/tests_new/mlops/test_model_versioning.py

@@ -32,7 +32,7 @@
 
 @pytest.fixture(scope="module")
 def reg_model1(winequality_vpy):
-    model = LinearRegression("reg_m1", solver="Newton", max_iter=2)
+    model = LinearRegression("reg_m1", solver="newton", max_iter=2)
     model.drop()
 
     model.fit(
@@ -46,7 +46,7 @@ def reg_model1(winequality_vpy):
 
 @pytest.fixture(scope="module")
 def reg_model2(winequality_vpy):
-    model = LinearRegression("reg_m2", solver="Newton", max_iter=2)
+    model = LinearRegression("reg_m2", solver="newton", max_iter=2)
     model.drop()
 
     model.fit(
@@ -60,7 +60,7 @@ def reg_model2(winequality_vpy):
 
 @pytest.fixture(scope="module")
 def reg_model3(winequality_vpy):
-    model = LinearRegression("reg_m3", solver="Newton", max_iter=2)
+    model = LinearRegression("reg_m3", solver="newton", max_iter=2)
     model.drop()
 
     model.fit(
@@ -74,7 +74,7 @@ def reg_model3(winequality_vpy):
 
 @pytest.fixture(scope="module")
 def bin_model1(winequality_vpy):
-    model = LogisticRegression("bin_m1", solver="Newton", max_iter=2, penalty=None)
+    model = LogisticRegression("bin_m1", solver="newton", max_iter=2, penalty=None)
     model.drop()
 
     model.fit(
@@ -88,7 +88,7 @@ def bin_model1(winequality_vpy):
 
 @pytest.fixture(scope="module")
 def bin_model2(winequality_vpy):
-    model = LogisticRegression("bin_m2", solver="Newton", max_iter=2, penalty=None)
+    model = LogisticRegression("bin_m2", solver="newton", max_iter=2, penalty=None)
     model.drop()
 
     model.fit(
@@ -110,7 +110,7 @@ class TestModelVersioning:
     def test_register_models(self, reg_model1):
         reg_model1.register("regression_app1", raise_error=True)
 
-        new_model = LinearRegression("new_m", solver="Newton", max_iter=5)
+        new_model = LinearRegression("new_m", solver="newton", max_iter=5)
         with pytest.raises(RuntimeError, match="Failed to register the model"):
             new_model.register("new_app", raise_error=True)
 

verticapy/tests_new/plotting/base_test_files.py

@@ -2130,7 +2130,7 @@ def plot_result(self, schema_loader, titanic_vd):
             name=f"{schema_loader}.test_LR_titanic",
             tol=1e-4,
             max_iter=100,
-            solver="Newton",
+            solver="newton",
         )
         stepwise_result = stepwise(
             model,
@@ -2190,7 +2190,7 @@ def test_additional_options_custom_height(self, titanic_vd):
         custom_height = 60
         custom_width = 70
         model = LogisticRegression(
-            name="test_LR_titanic", tol=1e-4, max_iter=100, solver="Newton"
+            name="test_LR_titanic", tol=1e-4, max_iter=100, solver="newton"
         )
         # Act
         stepwise_result = stepwise(