From 82136d2a7fe8f5a2c1177423f13312420cffe905 Mon Sep 17 00:00:00 2001
From: Prasanjit Prakash <jeet@gmail.com>
Date: Sat, 25 Feb 2017 16:31:22 +0530
Subject: [PATCH 1/3] PERF: categorical rank GH#15498

---
 pandas/core/algorithms.py  | 8 +++++---
 pandas/core/categorical.py | 7 +++++--
 2 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/pandas/core/algorithms.py b/pandas/core/algorithms.py
index b11927a80fb2e2..7f74067f7826b2 100644
--- a/pandas/core/algorithms.py
+++ b/pandas/core/algorithms.py
@@ -974,9 +974,6 @@ def _get_data_algo(values, func_map):
 
     f = None
 
-    if is_categorical_dtype(values):
-        values = values._values_for_rank()
-
     if is_float_dtype(values):
         f = func_map['float64']
         values = _ensure_float64(values)
@@ -992,6 +989,11 @@ def _get_data_algo(values, func_map):
     elif is_unsigned_integer_dtype(values):
         f = func_map['uint64']
         values = _ensure_uint64(values)
+
+    elif is_categorical_dtype(values):
+        f = func_map['float64']
+        values = values._values_for_rank()
+
     else:
         values = _ensure_object(values)
 
diff --git a/pandas/core/categorical.py b/pandas/core/categorical.py
index b88a6b171b316c..3a83a485c8c3bb 100644
--- a/pandas/core/categorical.py
+++ b/pandas/core/categorical.py
@@ -1416,14 +1416,17 @@ def _values_for_rank(self):
         numpy array
 
         """
+        from pandas import Series
         if self.ordered:
             values = self.codes
             mask = values == -1
+            values = values.astype('float64')
             if mask.any():
-                values = values.astype('float64')
                 values[mask] = np.nan
         else:
-            values = np.array(self)
+            values = np.array(
+                self.rename_categories(Series(self.categories).rank())
+            )
         return values
 
     def order(self, inplace=False, ascending=True, na_position='last'):

From 41e7b27ce9b4d62fac5a0c9e1f97a0f857e58aed Mon Sep 17 00:00:00 2001
From: Prasanjit Prakash <jeet@gmail.com>
Date: Mon, 27 Feb 2017 19:06:55 +0530
Subject: [PATCH 2/3]  PERF: categorical rank GH#15498

no need to rename categories where they are already ordered
---
 pandas/core/algorithms.py             |  7 +++----
 pandas/core/categorical.py            |  4 +++-
 pandas/tests/series/test_analytics.py | 10 ++++++++++
 3 files changed, 16 insertions(+), 5 deletions(-)

diff --git a/pandas/core/algorithms.py b/pandas/core/algorithms.py
index 7f74067f7826b2..55d404f05dd1d5 100644
--- a/pandas/core/algorithms.py
+++ b/pandas/core/algorithms.py
@@ -974,6 +974,9 @@ def _get_data_algo(values, func_map):
 
     f = None
 
+    if is_categorical_dtype(values):
+        values = values._values_for_rank()
+
     if is_float_dtype(values):
         f = func_map['float64']
         values = _ensure_float64(values)
@@ -990,10 +993,6 @@ def _get_data_algo(values, func_map):
         f = func_map['uint64']
         values = _ensure_uint64(values)
 
-    elif is_categorical_dtype(values):
-        f = func_map['float64']
-        values = values._values_for_rank()
-
     else:
         values = _ensure_object(values)
 
diff --git a/pandas/core/categorical.py b/pandas/core/categorical.py
index 3a83a485c8c3bb..a8ccae0d374cd1 100644
--- a/pandas/core/categorical.py
+++ b/pandas/core/categorical.py
@@ -1420,9 +1420,11 @@ def _values_for_rank(self):
         if self.ordered:
             values = self.codes
             mask = values == -1
-            values = values.astype('float64')
             if mask.any():
+                values = values.astype('float64')
                 values[mask] = np.nan
+        elif self.categories.is_monotonic:
+            values = np.array(self)
         else:
             values = np.array(
                 self.rename_categories(Series(self.categories).rank())
diff --git a/pandas/tests/series/test_analytics.py b/pandas/tests/series/test_analytics.py
index b092e4f0847670..1733c515b272ca 100644
--- a/pandas/tests/series/test_analytics.py
+++ b/pandas/tests/series/test_analytics.py
@@ -1083,6 +1083,16 @@ def test_rank_categorical(self):
         res = unordered.rank()
         assert_series_equal(res, exp_unordered)
 
+        unordered1 = pd.Series(
+            [1, 2, 3, 4, 5, 6],
+        ).astype('category').cat.set_categories(
+            [1, 2, 3, 4, 5, 6],
+            ordered=False
+        )
+        exp_unordered1 = pd.Series([1., 2., 3., 4., 5., 6.])
+        res1 = unordered1.rank()
+        assert_series_equal(res1, exp_unordered1)
+
         # Test na_option for rank data
         na_ser = pd.Series(
             ['first', 'second', 'third', 'fourth', 'fifth', 'sixth', np.NaN]

From f7eab32aed9d59547abba263cf084d29627e8290 Mon Sep 17 00:00:00 2001
From: Prasanjit Prakash <jeet@gmail.com>
Date: Mon, 27 Feb 2017 19:17:11 +0530
Subject: [PATCH 3/3]  PERF: categorical rank GH#15498

check for numeric instead of monotonic
---
 pandas/core/categorical.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pandas/core/categorical.py b/pandas/core/categorical.py
index a8ccae0d374cd1..2326cc3c78b72d 100644
--- a/pandas/core/categorical.py
+++ b/pandas/core/categorical.py
@@ -1423,7 +1423,7 @@ def _values_for_rank(self):
             if mask.any():
                 values = values.astype('float64')
                 values[mask] = np.nan
-        elif self.categories.is_monotonic:
+        elif self.categories.is_numeric():
             values = np.array(self)
         else:
             values = np.array(