FIX: add support for desc order when ranking infs with nans #19538

doc/source/whatsnew/v0.23.0.txt

@@ -1022,6 +1022,7 @@ Numeric
 - Bug in :class:`Index` constructor with ``dtype='uint64'`` where int-like floats were not coerced to :class:`UInt64Index` (:issue:`18400`)
 - Bug in :class:`DataFrame` flex arithmetic (e.g. ``df.add(other, fill_value=foo)``) with a ``fill_value`` other than ``None`` failed to raise ``NotImplementedError`` in corner cases where either the frame or ``other`` has length zero (:issue:`19522`)
 - Multiplication and division of numeric-dtyped :class:`Index` objects with timedelta-like scalars returns ``TimedeltaIndex`` instead of raising ``TypeError`` (:issue:`19333`)
+- Bug in :meth:`Series.rank` and :meth:`DataFrame.rank` when ``ascending='False'`` failed to return correct ranks for infinity if ``NaN`` were present (:issue:`19538`)
 - Bug where ``NaN`` was returned instead of 0 by :func:`Series.pct_change` and :func:`DataFrame.pct_change` when ``fill_method`` is not ``None`` (:issue:`19873`)
 
 

pandas/_libs/algos_rank_helper.pxi.in

@@ -135,7 +135,7 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
 
     sorted_data = values.take(_as)
     sorted_mask = mask.take(_as)
-    _indices = order[1].take(_as).nonzero()[0]
+    _indices = np.diff(sorted_mask).nonzero()[0]
     non_na_idx = _indices[0] if len(_indices) > 0 else -1
     argsorted = _as.astype('i8')
 
@@ -153,7 +153,7 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
 
         if (i == n - 1 or
                 are_diff(util.get_value_at(sorted_data, i + 1), val) or
-                i == non_na_idx - 1):
+                i == non_na_idx):
             if tiebreak == TIEBREAK_AVERAGE:
                 for j in range(i - dups + 1, i + 1):
                     ranks[argsorted[j]] = sum_ranks / dups
@@ -190,7 +190,7 @@ def rank_1d_{{dtype}}(object in_arr, ties_method='average', ascending=True,
             count += 1.0
 
             if (i == n - 1 or sorted_data[i + 1] != val or
-                i == non_na_idx - 1):
+                i == non_na_idx):
                 if tiebreak == TIEBREAK_AVERAGE:
                     for j in range(i - dups + 1, i + 1):
                         ranks[argsorted[j]] = sum_ranks / dups

pandas/tests/series/test_rank.py

@@ -16,6 +16,8 @@
 from pandas.tests.series.common import TestData
 from pandas._libs.tslib import iNaT
 from pandas._libs.algos import Infinity, NegInfinity
+from itertools import chain
+import pandas.util._test_decorators as td
 
 
 class TestSeriesRank(TestData):
@@ -257,38 +259,52 @@ def _check(s, expected, method='average'):
             series = s if dtype is None else s.astype(dtype)
             _check(series, results[method], method=method)
 
-    def test_rank_tie_methods_on_infs_nans(self):
+    @td.skip_if_no_scipy
+    @pytest.mark.parametrize('ascending', [True, False])
+    @pytest.mark.parametrize('method', ['average', 'min', 'max', 'first',
+                                        'dense'])
+    @pytest.mark.parametrize('na_option', ['top', 'bottom', 'keep'])
+    def test_rank_tie_methods_on_infs_nans(self, method, na_option, ascending):
         dtypes = [('object', None, Infinity(), NegInfinity()),
                   ('float64', np.nan, np.inf, -np.inf)]
         chunk = 3
         disabled = set([('object', 'first')])
 
-        def _check(s, expected, method='average', na_option='keep'):
-            result = s.rank(method=method, na_option=na_option)
+        def _check(s, method, na_option, ascending):
+            exp_ranks = {
+                'average': ([2, 2, 2], [5, 5, 5], [8, 8, 8]),
+                'min': ([1, 1, 1], [4, 4, 4], [7, 7, 7]),
+                'max': ([3, 3, 3], [6, 6, 6], [9, 9, 9]),
+                'first': ([1, 2, 3], [4, 5, 6], [7, 8, 9]),
+                'dense': ([1, 1, 1], [2, 2, 2], [3, 3, 3])
+            }
+            ranks = exp_ranks[method]
+            if na_option == 'top':
+                order = [ranks[1], ranks[0], ranks[2]]
+            elif na_option == 'bottom':
+                order = [ranks[0], ranks[2], ranks[1]]
+            else:
+                order = [ranks[0], [np.nan] * chunk, ranks[1]]
+            expected = order if ascending else order[::-1]
+            expected = list(chain.from_iterable(expected))
+            result = s.rank(method=method, na_option=na_option,
+                            ascending=ascending)
             tm.assert_series_equal(result, Series(expected, dtype='float64'))
 
-        exp_ranks = {
-            'average': ([2, 2, 2], [5, 5, 5], [8, 8, 8]),
-            'min': ([1, 1, 1], [4, 4, 4], [7, 7, 7]),
-            'max': ([3, 3, 3], [6, 6, 6], [9, 9, 9]),
-            'first': ([1, 2, 3], [4, 5, 6], [7, 8, 9]),
-            'dense': ([1, 1, 1], [2, 2, 2], [3, 3, 3])
-        }
-        na_options = ('top', 'bottom', 'keep')
         for dtype, na_value, pos_inf, neg_inf in dtypes:
             in_arr = [neg_inf] * chunk + [na_value] * chunk + [pos_inf] * chunk
             iseries = Series(in_arr, dtype=dtype)
-            for method, na_opt in product(exp_ranks.keys(), na_options):
-                ranks = exp_ranks[method]
-                if (dtype, method) in disabled:
-                    continue
-                if na_opt == 'top':
-                    order = ranks[1] + ranks[0] + ranks[2]
-                elif na_opt == 'bottom':
-                    order = ranks[0] + ranks[2] + ranks[1]
-                else:
-                    order = ranks[0] + [np.nan] * chunk + ranks[1]
-                _check(iseries, order, method, na_opt)
+            if (dtype, method) in disabled:
+                continue
+            _check(iseries, method, na_option, ascending)
+
+    def test_rank_desc_mix_nans_infs(self):
+        # GH 19538
+        # check descending ranking when mix nans and infs
+        iseries = Series([1, np.nan, np.inf, -np.inf, 25])
+        result = iseries.rank(ascending=False)
+        exp = Series([3, np.nan, 1, 4, 2], dtype='float64')
+        tm.assert_series_equal(result, exp)
 
     def test_rank_methods_series(self):
         pytest.importorskip('scipy.stats.special')