Merge pull request #6459 from qwhelan/set_index_names

ENH: Preserve .names in df.set_index(df.index)

Author: jreback

doc/source/release.rst

@@ -107,6 +107,7 @@ API Changes
   or numbering columns as needed (:issue:`2385`)
 - Slicing and advanced/boolean indexing operations on ``Index`` classes will no
   longer change type of the resulting index (:issue:`6440`).
+- ``set_index`` no longer converts MultiIndexes to an Index of tuples (:issue:`6459`).
 
 Experimental Features
 ~~~~~~~~~~~~~~~~~~~~~
@@ -197,6 +198,7 @@ Bug Fixes
 - Bug in multi-axis indexing using ``.loc`` on non-unique indices (:issue:`6504`)
 - Bug that caused _ref_locs corruption when slice indexing across columns axis of a DataFrame (:issue:`6525`)
 - Regression from 0.13 in the treatmenet of numpy ``datetime64`` non-ns dtypes in Series creation (:issue:`6529`)
+- ``.names`` attribute of MultiIndexes passed to ``set_index`` are now preserved (:issue:`6459`).
 
 pandas 0.13.1
 -------------

doc/source/v0.14.0.txt

@@ -92,6 +92,49 @@ These are out-of-bounds selections
   .. ipython:: python
 
      i[[0,1,2]].astype(np.int_)
+- ``set_index`` no longer converts MultiIndexes to an Index of tuples. For example,
+  the old behavior returned an Index in this case (:issue:`6459`):
+
+  .. ipython:: python
+     :suppress:
+
+     from itertools import product
+     tuples = list(product(('a', 'b'), ('c', 'd')))
+     mi = MultiIndex.from_tuples(tuples)
+     df_multi = DataFrame(np.random.randn(4, 2), index=mi)
+     tuple_ind = pd.Index(tuples)
+
+  .. ipython:: python
+
+     df_multi.index
+
+     @suppress
+     df_multi.index = tuple_ind
+
+     # Old behavior, casted MultiIndex to an Index
+     df_multi.set_index(df_multi.index)
+
+     @suppress
+     df_multi.index = mi
+
+     # New behavior
+     df_multi.set_index(df_multi.index)
+
+  This also applies when passing multiple indices to ``set_index``:
+
+  .. ipython:: python
+
+    @suppress
+    df_multi.index = tuple_ind
+
+    # Old output, 2-level MultiIndex of tuples
+    df_multi.set_index([df_multi.index, df_multi.index])
+
+    @suppress
+    df_multi.index = mi
+
+    # New output, 4-level MultiIndex
+    df_multi.set_index([df_multi.index, df_multi.index])
 
 
 MultiIndexing Using Slicers

pandas/core/frame.py

@@ -2240,7 +2240,15 @@ def set_index(self, keys, drop=True, append=False, inplace=False,
 
         to_remove = []
         for col in keys:
-            if isinstance(col, Series):
+            if isinstance(col, MultiIndex):
+                # append all but the last column so we don't have to modify
+                # the end of this loop
+                for n in range(col.nlevels - 1):
+                    arrays.append(col.get_level_values(n))
+
+                level = col.get_level_values(col.nlevels - 1)
+                names.extend(col.names)
+            elif isinstance(col, (Series, Index)):
                 level = col.values
                 names.append(col.name)
             elif isinstance(col, (list, np.ndarray)):

pandas/tests/test_frame.py

@@ -12280,6 +12280,33 @@ def test_dtypes_are_correct_after_column_slice(self):
                             pd.Series(odict([('a', np.float_), ('b', np.float_),
                                              ('c', np.float_),])))
 
+    def test_set_index_names(self):
+        df = pd.util.testing.makeDataFrame()
+        df.index.name = 'name'
+
+        self.assertEquals(df.set_index(df.index).index.names, ['name'])
+
+        mi = MultiIndex.from_arrays(df[['A', 'B']].T.values, names=['A', 'B'])
+        mi2 = MultiIndex.from_arrays(df[['A', 'B', 'A', 'B']].T.values,
+                                     names=['A', 'B', 'A', 'B'])
+
+        df = df.set_index(['A', 'B'])
+
+        self.assertEquals(df.set_index(df.index).index.names, ['A', 'B'])
+
+        # Check that set_index isn't converting a MultiIndex into an Index
+        self.assertTrue(isinstance(df.set_index(df.index).index, MultiIndex))
+
+        # Check actual equality
+        tm.assert_index_equal(df.set_index(df.index).index, mi)
+
+        # Check that [MultiIndex, MultiIndex] yields a MultiIndex rather
+        # than a pair of tuples
+        self.assertTrue(isinstance(df.set_index([df.index, df.index]).index, MultiIndex))
+
+        # Check equality
+        tm.assert_index_equal(df.set_index([df.index, df.index]).index, mi2)
+
 
 def skip_if_no_ne(engine='numexpr'):
     if engine == 'numexpr':