BUG: boxplot fails when one column is all NaNs

Add test case for empty column Replace empty arrays with [np.nan] Add fix to the release notes Add issue numbers

Author: onesandzeroes

doc/source/v0.15.0.txt

@@ -616,8 +616,8 @@ Enhancements
 
  .. code-block:: python
 
- df.to_sql('table', engine, schema='other_schema')
- pd.read_sql_table('table', engine, schema='other_schema')
+  df.to_sql('table', engine, schema='other_schema')
+  pd.read_sql_table('table', engine, schema='other_schema')
 
 - Added support for writing ``NaN`` values with ``to_sql`` (:issue:`2754`).
 - Added support for writing datetime64 columns with ``to_sql`` for all database flavors (:issue:`7103`).
@@ -919,3 +919,4 @@ Bug Fixes
 - Bug with ``DatetimeIndex.asof`` incorrectly matching partial strings and returning the wrong date (:issue:`8245`).
 - Bug in plotting methods modifying the global matplotlib rcParams (:issue:`8242`).
 - Bug in ``DataFrame.__setitem__`` that caused errors when setting a dataframe column to a sparse array (:issue:`8131`)
+- Bug where ``Dataframe.boxplot()`` failed when entire column was empty (:issue:`8181`).

pandas/tests/test_graphics.py

@@ -1915,6 +1915,11 @@ def _check_ax_limits(col, ax):
  self.assertEqual(age_ax._sharey, height_ax)
  self.assertIsNone(dummy_ax._sharey)
 
+ @slow
+ def test_boxplot_empty_column(self):
+ df = DataFrame(np.random.randn(20, 4))
+ df.loc[:, 0] = np.nan
+ _check_plot_works(df.boxplot, return_type='axes')
 
  @slow
  def test_kde_df(self):

pandas/tools/plotting.py

@@ -2081,6 +2081,10 @@ def _get_plot_function(self):
  def plotf(ax, y, column_num=None, **kwds):
  if y.ndim == 2:
  y = [remove_na(v) for v in y]
+ # Boxplot fails with empty arrays, so need to add a NaN
+ # if any cols are empty
+ # GH 8181
+ y = [v if v.size > 0 else np.array([np.nan]) for v in y]
  else:
  y = remove_na(y)
  bp = ax.boxplot(y, **kwds)