PERF: rework MultiIndex.is_monotonic as per @ssanderson idea

Author: jreback

asv_bench/benchmarks/indexing.py

@@ -189,7 +189,9 @@ def setup(self):
  self.eps_C = 5
  self.eps_D = 5000
  self.mdt2 = self.mdt.set_index(['A', 'B', 'C', 'D']).sortlevel()
- self.miint = MultiIndex.from_product([np.arange(1000), np.arange(1000)], names=['one','two'])
+ self.miint = MultiIndex.from_product(
+ [np.arange(1000),
+ np.arange(1000)], names=['one', 'two'])
 
  def time_series_xs_mi_ix(self):
  self.s.ix[999]
@@ -198,11 +200,21 @@ def time_frame_xs_mi_ix(self):
  self.df.ix[999]
 
  def time_multiindex_slicers(self):
- self.mdt2.loc[self.idx[(self.test_A - self.eps_A):(self.test_A + self.eps_A), (self.test_B - self.eps_B):(self.test_B + self.eps_B), (self.test_C - self.eps_C):(self.test_C + self.eps_C), (self.test_D - self.eps_D):(self.test_D + self.eps_D)], :]
+ self.mdt2.loc[self.idx[
+ (self.test_A - self.eps_A):(self.test_A + self.eps_A),
+ (self.test_B - self.eps_B):(self.test_B + self.eps_B),
+ (self.test_C - self.eps_C):(self.test_C + self.eps_C),
+ (self.test_D - self.eps_D):(self.test_D + self.eps_D)], :]
 
  def time_multiindex_get_indexer(self):
- self.miint.get_indexer(np.array([(0, 10), (0, 11), (0, 12), (0, 13), (0, 14),
- (0, 15), (0, 16),(0, 17), (0, 18), (0, 19)], dtype=object))
+ self.miint.get_indexer(
+ np.array([(0, 10), (0, 11), (0, 12),
+ (0, 13), (0, 14), (0, 15),
+ (0, 16), (0, 17), (0, 18),
+ (0, 19)], dtype=object))
+
+ def time_is_monotonic(self):
+ self.miint.is_monotonic
 
 
 class PanelIndexing(object):

pandas/indexes/multi.py

@@ -667,15 +667,21 @@ def _has_complex_internals(self):
  @cache_readonly
  def is_monotonic(self):
 
- # TODO
- # this is unfortunate we end up tupelizing
- # just to determine monotonicity :<
-
- # fast-path
- if not self.levels[0].is_monotonic:
- return False
+ def level_values(level):
+ unique = self.levels[level]
+ labels = self.labels[level]
+ return algos.take_1d(unique.values, labels,
+ fill_value=unique._na_value)
+
+ # reversed() because lexsort() wants the most significant key last.
+ values = [level_values(i) for i in reversed(range(len(self.levels)))]
+ try:
+ sort_order = np.lexsort(values)
+ return Index(sort_order).is_monotonic
+ except TypeError:
 
- return Index(self.values).is_monotonic
+ # we have mixed types and np.lexsort is not happy
+ return Index(self.values).is_monotonic
 
  @cache_readonly
  def is_unique(self):

pandas/tests/indexes/test_multi.py

@@ -2318,11 +2318,50 @@ def test_level_setting_resets_attributes(self):
  ind = MultiIndex.from_arrays([
  ['A', 'A', 'B', 'B', 'B'], [1, 2, 1, 2, 3]
  ])
- assert ind.is_monotonic
+ self.assertTrue(ind.is_monotonic)
  ind.set_levels([['A', 'B', 'A', 'A', 'B'], [2, 1, 3, -2, 5]],
  inplace=True)
+
  # if this fails, probably didn't reset the cache correctly.
- assert not ind.is_monotonic
+ self.assertFalse(ind.is_monotonic)
+
+ def test_is_monotonic(self):
+ i = MultiIndex.from_product([np.arange(10),
+ np.arange(10)], names=['one', 'two'])
+ self.assertTrue(i.is_monotonic)
+ self.assertTrue(Index(i.values).is_monotonic)
+
+ i = MultiIndex.from_product([np.arange(10, 0, -1),
+ np.arange(10)], names=['one', 'two'])
+ self.assertFalse(i.is_monotonic)
+ self.assertFalse(Index(i.values).is_monotonic)
+
+ i = MultiIndex.from_product([np.arange(10),
+ np.arange(10, 0, -1)],
+ names=['one', 'two'])
+ self.assertFalse(i.is_monotonic)
+ self.assertFalse(Index(i.values).is_monotonic)
+
+ i = MultiIndex.from_product([[1.0, np.nan, 2.0], ['a', 'b', 'c']])
+ self.assertFalse(i.is_monotonic)
+ self.assertFalse(Index(i.values).is_monotonic)
+
+ # string ordering
+ i = MultiIndex(levels=[['foo', 'bar', 'baz', 'qux'],
+ ['one', 'two', 'three']],
+ labels=[[0, 0, 0, 1, 1, 2, 2, 3, 3, 3],
+ [0, 1, 2, 0, 1, 1, 2, 0, 1, 2]],
+ names=['first', 'second'])
+ self.assertFalse(i.is_monotonic)
+ self.assertFalse(Index(i.values).is_monotonic)
+
+ i = MultiIndex(levels=[['bar', 'baz', 'foo', 'qux'],
+ ['mom', 'next', 'zenith']],
+ labels=[[0, 0, 0, 1, 1, 2, 2, 3, 3, 3],
+ [0, 1, 2, 0, 1, 1, 2, 0, 1, 2]],
+ names=['first', 'second'])
+ self.assertTrue(i.is_monotonic)
+ self.assertTrue(Index(i.values).is_monotonic)
 
  def test_isin(self):
  values = [('foo', 2), ('bar', 3), ('quux', 4)]