Remove boxcox (#177)

* Turn off BoxCox transformer because it may lead to potential bugs

* change the decimal point comparison due to the randomness in TPOT

foreshadow/smart/all.py

@@ -22,7 +22,6 @@
     TfidfVectorizer,
 )
 from foreshadow.concrete.internals import (
-    BoxCox,
     ConvertFinancial,
     DummyEncoder,
     FancyImputer,
@@ -85,7 +84,14 @@ def pick_transformer(self, X, y=None, **fit_params):
         best_dist = best_dist if p_vals[best_dist] >= self.p_val else None
         if best_dist is None:
             return SerializablePipeline(
-                [("box_cox", BoxCox()), ("robust_scaler", RobustScaler())]
+                [
+                    # Turning off the BoxCox transformer because if the test
+                    # dataset has an even smaller negative min, it will
+                    # break the pipeline.
+                    # TODO add a different transformer if necessary
+                    # ("box_cox", BoxCox()),
+                    ("robust_scaler", RobustScaler())
+                ]
             )
         else:
             return distributions[best_dist]

foreshadow/tests/test_foreshadow.py

@@ -847,7 +847,7 @@ def test_foreshadow_serialization_adults_small_classification():
     # converge. The test here aims to evaluate if both cases have
     # produced a reasonable score and the difference is small.
     # assert score1 > 0.76 and score2 > 0.76
-    assertions.assertAlmostEqual(score1, score2, places=3)
+    assertions.assertAlmostEqual(score1, score2, places=2)
 
 
 def test_foreshadow_serialization_adults_small_classification_override():

foreshadow/tests/test_transformers/test_concrete/test_internals/test_internal.py

@@ -61,23 +61,6 @@ def test_dummy_encoder_other():
     assert check.equals(df)
 
 
-def test_box_cox():
-    import numpy as np
-    import pandas as pd
-    import scipy.stats as ss
-
-    from foreshadow.concrete import BoxCox
-
-    np.random.seed(0)
-    data = pd.DataFrame(ss.lognorm.rvs(size=100, s=0.954))
-    bc = BoxCox()
-    bc_data = bc.fit_transform(data)
-    assert ss.shapiro(bc_data)[1] > 0.05
-    assert np.allclose(
-        data.values.ravel(), bc.inverse_transform(bc_data).values.ravel()
-    )
-
-
 @pytest.mark.parametrize("deep", [True, False])
 def test_label_encoder_get_params_keys(deep):
     """Test that the desired keys show up for the LabelEncoder object.

foreshadow/tests/test_transformers/test_smart/test_smart.py

@@ -50,9 +50,7 @@ def test_smart_get_params_default(smart_child, smart_params, deep):
     assert smart_params == params
 
 
-@pytest.mark.parametrize(
-    "initial_transformer", [None, "BoxCox", "StandardScaler"]
-)
+@pytest.mark.parametrize("initial_transformer", [None, "StandardScaler"])
 def test_smart_get_params_deep(smart_child, smart_params, initial_transformer):
     """Test that smart.get_params(deep=True) functions as desired.
 
@@ -78,9 +76,7 @@ def test_smart_get_params_deep(smart_child, smart_params, initial_transformer):
     assert smart.get_params(True) == smart_params
 
 
-@pytest.mark.parametrize(
-    "initial_transformer", [None, "BoxCox", "StandardScaler"]
-)
+@pytest.mark.parametrize("initial_transformer", [None, "StandardScaler"])
 def test_smart_set_params_default(smart_child, initial_transformer):
     """Test setting both transformer and its parameters simultaneously works.