ENH: Add warning when setting into nonexistent attribute

doc/source/indexing.rst

@@ -227,10 +227,6 @@ as an attribute:
    dfa.A
    panel.one
 
-You can use attribute access to modify an existing element of a Series or column of a DataFrame, but be careful;
-if you try to use attribute access to create a new column, it fails silently, creating a new attribute rather than a
-new column.
-
 .. ipython:: python
 
    sa.a = 5
@@ -267,6 +263,37 @@ You can also assign a ``dict`` to a row of a ``DataFrame``:
    x.iloc[1] = dict(x=9, y=99)
    x
 
+You can use attribute access to modify an existing element of a Series or column of a DataFrame, but be careful;
+if you try to use attribute access to create a new column, it creates a new attribute rather than a
+new column. In 0.21.0 and later, this will raise a ``UserWarning``:
+
+.. code-block:: ipython
+
+    In[1]: df = pd.DataFrame({'one': [1., 2., 3.]})
+    In[2]: df.two = [4, 5, 6]
+    UserWarning: Pandas doesn't allow Series to be assigned into nonexistent columns - see https://pandas.pydata.org/pandas-docs/stable/indexing.html#attribute_access
+    In[3]: df
+    Out[3]:
+       one
+    0  1.0
+    1  2.0
+    2  3.0
+
+Similarly, it is possible to create a column with a name which collides with one of Pandas's
+built-in methods or attributes, which can cause confusion later when attempting to access
+that column as an attribute. This behavior now warns:
+
+.. code-block:: ipython
+
+    In[4]: df['sum'] = [5., 7., 9.]
+    UserWarning: Column name 'sum' collides with a built-in method, which will cause unexpected attribute behavior
+    In[5]: df.sum
+    Out[5]:
+    <bound method DataFrame.sum of    one  sum
+    0  1.0  5.0
+    1  2.0  7.0
+    2  3.0  9.0>
+
 Slicing ranges
 --------------
 

doc/source/whatsnew/v0.21.0.txt

@@ -29,7 +29,6 @@ New features
 - Added ``skipna`` parameter to :func:`~pandas.api.types.infer_dtype` to
   support type inference in the presence of missing values (:issue:`17059`).
 
-
 .. _whatsnew_0210.enhancements.infer_objects:
 
 ``infer_objects`` type conversion
@@ -62,6 +61,51 @@ using the :func:`to_numeric` function (or :func:`to_datetime`, :func:`to_timedel
    df['C'] = pd.to_numeric(df['C'], errors='coerce')
    df.dtypes
 
+.. _whatsnew_0210.enhancements.attribute_access:
+
+Improved warnings when attempting to create columns
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+New users are often flummoxed by the relationship between column operations and attribute
+access on ``DataFrame`` instances (:issue:`5904` & :issue:`7175`). Two specific instances
+of this confusion include attempting to create a new column by setting into an attribute:
+
+.. code-block:: ipython
+
+  In[1]: df = pd.DataFrame({'one': [1., 2., 3.]})
+  In[2]: df.two = [4, 5, 6]
+
+This does not raise any obvious exceptions, but also does not create a new column:
+
+.. code-block:: ipython
+
+  In[3]: df
+  Out[3]:
+      one
+  0  1.0
+  1  2.0
+  2  3.0
+
+The second source of confusion is creating a column whose name collides with a method or
+attribute already in the instance namespace:
+
+.. code-block:: ipython
+
+  In[4]: df['sum'] = [5., 7., 9.]
+
+This does not permit that column to be accessed as an attribute:
+
+.. code-block:: ipython
+
+  In[5]: df.sum
+  Out[5]:
+  <bound method DataFrame.sum of    one  sum
+  0  1.0  5.0
+  1  2.0  7.0
+  2  3.0  9.0>
+
+Both of these now raise a ``UserWarning`` about the potential for unexpected behavior. See :ref:`Attribute Access <indexing.attribute_access>`.
+
 .. _whatsnew_0210.enhancements.other:
 
 Other Enhancements

pandas/core/generic.py

@@ -27,7 +27,7 @@
     pandas_dtype)
 from pandas.core.dtypes.cast import maybe_promote, maybe_upcast_putmask
 from pandas.core.dtypes.missing import isna, notna
-from pandas.core.dtypes.generic import ABCSeries, ABCPanel
+from pandas.core.dtypes.generic import ABCSeries, ABCPanel, ABCDataFrame
 
 from pandas.core.common import (_values_from_object,
                                 _maybe_box_datetimelike,
@@ -1907,6 +1907,10 @@ def _slice(self, slobj, axis=0, kind=None):
         return result
 
     def _set_item(self, key, value):
+        if isinstance(key, str) and callable(getattr(self, key, None)):
+            warnings.warn("Column name '{key}' collides with a built-in "
+                          "method, which will cause unexpected attribute "
+                          "behavior".format(key=key), stacklevel=3)
         self._data.set(key, value)
         self._clear_item_cache()
 
@@ -3357,6 +3361,12 @@ def __setattr__(self, name, value):
                 else:
                     object.__setattr__(self, name, value)
             except (AttributeError, TypeError):
+                if isinstance(self, ABCDataFrame) and (is_list_like(value)):
+                    warnings.warn("Pandas doesn't allow Series to be assigned "
+                                  "into nonexistent columns - see "
+                                  "https://pandas.pydata.org/pandas-docs/"
+                                  "stable/indexing.html#attribute-access",
+                                  stacklevel=2)
                 object.__setattr__(self, name, value)
 
     # ----------------------------------------------------------------------

pandas/tests/dtypes/test_generic.py

@@ -4,6 +4,7 @@
 import numpy as np
 import pandas as pd
 from pandas.core.dtypes import generic as gt
+from pandas.util import testing as tm
 
 
 class TestABCClasses(object):
@@ -38,3 +39,40 @@ def test_abc_types(self):
         assert isinstance(self.sparse_array, gt.ABCSparseArray)
         assert isinstance(self.categorical, gt.ABCCategorical)
         assert isinstance(pd.Period('2012', freq='A-DEC'), gt.ABCPeriod)
+
+
+def test_setattr_warnings():
+    # GH5904 - Suggestion: Warning for DataFrame colname-methodname clash
+    # GH7175 - GOTCHA: You can't use dot notation to add a column...
+    d = {'one': pd.Series([1., 2., 3.], index=['a', 'b', 'c']),
+         'two': pd.Series([1., 2., 3., 4.], index=['a', 'b', 'c', 'd'])}
+    df = pd.DataFrame(d)
+
+    with catch_warnings(record=True) as w:
+        #  successfully add new column
+        #  this should not raise a warning
+        df['three'] = df.two + 1
+        assert len(w) == 0
+        assert df.three.sum() > df.two.sum()
+
+    with catch_warnings(record=True) as w:
+        #  successfully modify column in place
+        #  this should not raise a warning
+        df.one += 1
+        assert len(w) == 0
+        assert df.one.iloc[0] == 2
+
+    with catch_warnings(record=True) as w:
+        #  successfully add an attribute to a series
+        #  this should not raise a warning
+        df.two.not_an_index = [1, 2]
+        assert len(w) == 0
+
+    with tm.assert_produces_warning(UserWarning):
+        #  warn when setting column to nonexistent name
+        df.four = df.two + 2
+        assert df.four.sum() > df.two.sum()
+
+    with tm.assert_produces_warning(UserWarning):
+        #  warn when column has same name as method
+        df['sum'] = df.two

pandas/tests/io/test_pytables.py

@@ -2011,7 +2011,7 @@ def check(obj, comparator):
         df['string'] = 'foo'
         df['float322'] = 1.
         df['float322'] = df['float322'].astype('float32')
-        df['bool'] = df['float322'] > 0
+        df['boolean'] = df['float322'] > 0
         df['time1'] = Timestamp('20130101')
         df['time2'] = Timestamp('20130102')
         check(df, tm.assert_frame_equal)
@@ -2141,7 +2141,7 @@ def test_table_values_dtypes_roundtrip(self):
             df1['string'] = 'foo'
             df1['float322'] = 1.
             df1['float322'] = df1['float322'].astype('float32')
-            df1['bool'] = df1['float32'] > 0
+            df1['boolean'] = df1['float32'] > 0
             df1['time1'] = Timestamp('20130101')
             df1['time2'] = Timestamp('20130102')