add entry to whatsnew

Author: peterpanmj

doc/source/whatsnew/v0.22.0.txt

@@ -22,6 +22,87 @@ New features
 Other Enhancements
 ^^^^^^^^^^^^^^^^^^
 
+correctly calculate ranks for infinit values
+""""""""""""""""""""""""""""""""""""""""""""
+
+In previous versions, ``inf`` elements were assigned ``nan`` as their ranks. Now ranks are calculated properly.
+
+Previous Behavior:
+
+.. code-block:: ipython
+
+    In [17]: pd.Series([-np.inf, 0, 1, np.inf]).rank()
+    Out[17]:
+    0    1.0
+    1    2.0
+    2    3.0
+    3    NaN
+
+Current Behavior
+
+.. code-block:: ipython
+
+    In [5]:  pd.Series([-np.inf, 0, 1, np.inf]).rank()
+    Out[5]:
+    0    1.0
+    1    2.0
+    2    3.0
+    3    4.0
+    dtype: float64
+
+Furthermore, previously if you rank ``inf`` or ``-inf`` values together with ``nan`` values, results were wrong when using 'top' or 'bottom' argument.
+
+Previously Behavior:
+
+.. code-block:: ipython
+
+    In [15]: pd.Series([np.nan, np.nan, -np.inf, -np.inf]).rank(na_option='top')
+    Out[15]:
+    0    2.5
+    1    2.5
+    2    2.5
+    3    2.5
+    dtype: float64
+
+Current Behavior
+
+.. code-block:: ipython
+
+    In [4]:  pd.Series([np.nan, np.nan, -np.inf, -np.inf]).rank(na_option='top')
+    Out[4]:
+    0    1.5
+    1    1.5
+    2    3.5
+    3    3.5
+    dtype: float64
+
+Moreover, previously if you rank an array of `object` dtype, ``None`` values will have different ranks.
+
+Previously Behavior:
+
+.. code-block:: ipython
+
+    In [3]:  pd.Series([None, None, None, 'A','B']).rank(na_option='top')
+    Out[3]:
+    0    3.0
+    1    2.0
+    2    1.0
+    3    4.0
+    4    5.0
+    dtype: float64
+
+Current Behavior
+
+.. code-block:: ipython
+
+    In [3]:  pd.Series([None, None, None, 'A','B']).rank(na_option='top')
+    Out[3]:
+    0    2.0
+    1    2.0
+    2    2.0
+    3    4.0
+    4    5.0
+
 - Better support for ``Dataframe.style.to_excel()`` output with the ``xlsxwriter`` engine. (:issue:`16149`)
 -
 -
@@ -142,7 +223,7 @@ Reshaping
 Numeric
 ^^^^^^^
 
--
+- Bug in ``pd.Series.rank()`` and ``pd.DataFrame.rank() could not properly rank infinit values. Infinit values were assigned NaNs as ranks. If NaNs were present together with infinit values, the ranks were calculated wrong (:issue:`6945`)
 -
 -