Merge pull request #23 from jnothman/pr3886

amueller · amueller · commit 0a3b75a45d1c · 2014-12-24T23:23:53.000-05:00
Some finishing touches to the FeatureUnion changes
diff --git a/doc/modules/pipeline.rst b/doc/modules/pipeline.rst
@@ -122,9 +122,11 @@ FeatureUnion: composite feature spaces
 :class:`FeatureUnion` combines several transformer objects into a new
 transformer that combines their output. A :class:`FeatureUnion` takes
 a list of transformer objects. During fitting, each of these
-is fit to the data independently. For transforming data, the
-transformers are applied in parallel, and the sample vectors they output
-are concatenated end-to-end into larger vectors.
+is fit to the data independently. It can also be used to apply different
+transformations to each field of the data, producing a homogeneous feature
+matrix from a heterogeneous data source.
+The transformers are applied in parallel, and the feature matrices they output
+are concatenated side-by-side into a larger matrix.
 
 :class:`FeatureUnion` serves the same purposes as :class:`Pipeline` -
 convenience and joint parameter estimation and validation.
@@ -166,4 +168,4 @@ Like pipelines, feature unions have a shorthand constructor called
 .. topic:: Examples:
 
  * :ref:`example_feature_stacker.py`
- * :ref:`example_hetero_feature_union.py`
+ * :ref:`example_hetero_feature_union.py` illustrates the ``fields`` parameter.
diff --git a/sklearn/pipeline.py b/sklearn/pipeline.py
@@ -292,8 +292,9 @@ class FeatureUnion(BaseEstimator, TransformerMixin):
     """Concatenates results of multiple transformer objects.
 
     This estimator applies a list of transformer objects in parallel to the
-    input data, then concatenates the results. This is useful to combine
-    several feature extraction mechanisms into a single transformer.
+    input data, or each to a different field of the input, then concatenates
+    the results. This is useful to combine several feature extraction
+    mechanisms into a single transformer.
 
     Parameters
     ----------
@@ -415,7 +416,8 @@ def transform(self, X):
         Xs = Parallel(n_jobs=self.n_jobs)(
             delayed(_transform_one)(trans, name, X, self.transformer_weights,
                                     field)
-            for (name, trans), field in zip(self.transformer_list, self.fields_))
+            for (name, trans), field in zip(self.transformer_list,
+                                            self.fields_))
         if any(sparse.issparse(f) for f in Xs):
             Xs = sparse.hstack(Xs).tocsr()
         else:
@@ -443,8 +445,8 @@ def _check_fields(self):
             fields = self.fields
             if len(fields) != len(self.transformer_list):
                 raise ValueError("Length of transformer list %d does not match"
-                                 " length of fields %d" % (len(self.transformer_list),
-                                                           len(fields)))
+                                 " length of fields %d" %
+                                 (len(self.transformer_list), len(fields)))
         else:
             fields = [None for x in self.transformer_list]
         return fields
