Merge pull request #19 from georgianpartners/issue_17

Change np.isnan to pd.isnull to get around string error, issue #17

foreshadow/tests/test_transformers/test_internal.py

@@ -119,3 +119,16 @@ def test_drop_transformer_below_thresh():
     x = pd.DataFrame({"A": np.array([np.nan] * 8 + [0.1, 0.1])})
 
     assert DropFeature().fit_transform(x).values.size == 0
+
+
+def test_drop_transformer_string_input():
+    import uuid
+    import numpy as np
+    import pandas as pd
+    from foreshadow.transformers.internals import DropFeature
+
+    x = pd.DataFrame({"A": np.array([str(uuid.uuid4()) for _ in range(40)])})
+
+    assert np.array_equal(
+        x.values.ravel(), DropFeature().fit_transform(x).values.ravel()
+    )

foreshadow/transformers/internals/dropfeature.py

@@ -1,4 +1,5 @@
 import numpy as np
+import pandas as pd
 from sklearn.base import TransformerMixin, BaseEstimator
 from sklearn.utils import check_array
 from sklearn.utils.validation import check_is_fitted
@@ -22,8 +23,9 @@ def fit(self, X, y=None):
             self
 
         """
-        X = check_array(X, force_all_finite=False).ravel()
-        ratio = np.count_nonzero(np.isfinite(X)) / X.size
+        X = check_array(X, force_all_finite=False, dtype=None).ravel()
+        ratio = np.count_nonzero(~pd.isnull(X)) / X.size
+
         self.drop_ = ratio < self.threshold
         return self
 
@@ -37,6 +39,6 @@ def transform(self, X, y=None):
             :obj:`pandas.DataFrame`: Transformed data
 
         """
-        X = check_array(X, force_all_finite=False, copy=True)
+        X = check_array(X, force_all_finite=False, dtype=None, copy=True)
         check_is_fitted(self, "drop_")
         return X if not self.drop_ else np.array([])