Skip to content

Commit da03710

Browse files
sinhrksjreback
sinhrks
authored and
jreback
committed
TST: Add more Sparse indexing tests
closes #4400 Author: sinhrks <[email protected]> Closes #12848 from sinhrks/sparse_test2 and squashes the following commits: bb479e3 [sinhrks] TST: Add more Sparse indexing tests
1 parent a1f5ef3 commit da03710

File tree

2 files changed

+235
-1
lines changed

2 files changed

+235
-1
lines changed

‎pandas/sparse/tests/test_indexing.py

+234
Original file line numberDiff line numberDiff line change
@@ -32,6 +32,14 @@ def test_getitem(self):
3232
exp = orig[orig % 2 == 1].to_sparse()
3333
tm.assert_sp_series_equal(result, exp)
3434

35+
def test_getitem_slice(self):
36+
orig = pd.Series([1, np.nan, np.nan, 3, np.nan])
37+
sparse = orig.to_sparse()
38+
tm.assert_sp_series_equal(sparse[:2], orig[:2].to_sparse())
39+
tm.assert_sp_series_equal(sparse[4:2], orig[4:2].to_sparse())
40+
tm.assert_sp_series_equal(sparse[::2], orig[::2].to_sparse())
41+
tm.assert_sp_series_equal(sparse[-5:], orig[-5:].to_sparse())
42+
3543
def test_getitem_fill_value(self):
3644
orig = pd.Series([1, np.nan, 0, 3, 0])
3745
sparse = orig.to_sparse(fill_value=0)
@@ -63,6 +71,18 @@ def test_getitem_ellipsis(self):
6371
s = pd.SparseSeries([1, np.nan, 2, 0, np.nan], fill_value=0)
6472
tm.assert_sp_series_equal(s[...], s)
6573

74+
def test_getitem_slice_fill_value(self):
75+
orig = pd.Series([1, np.nan, 0, 3, 0])
76+
sparse = orig.to_sparse(fill_value=0)
77+
tm.assert_sp_series_equal(sparse[:2],
78+
orig[:2].to_sparse(fill_value=0))
79+
tm.assert_sp_series_equal(sparse[4:2],
80+
orig[4:2].to_sparse(fill_value=0))
81+
tm.assert_sp_series_equal(sparse[::2],
82+
orig[::2].to_sparse(fill_value=0))
83+
tm.assert_sp_series_equal(sparse[-5:],
84+
orig[-5:].to_sparse(fill_value=0))
85+
6686
def test_loc(self):
6787
orig = pd.Series([1, np.nan, np.nan, 3, np.nan])
6888
sparse = orig.to_sparse()
@@ -237,6 +257,25 @@ def test_iat_fill_value(self):
237257
self.assertEqual(sparse.iat[-1], orig.iat[-1])
238258
self.assertEqual(sparse.iat[-5], orig.iat[-5])
239259

260+
def test_get(self):
261+
s = pd.SparseSeries([1, np.nan, np.nan, 3, np.nan])
262+
self.assertEqual(s.get(0), 1)
263+
self.assertTrue(np.isnan(s.get(1)))
264+
self.assertIsNone(s.get(5))
265+
266+
s = pd.SparseSeries([1, np.nan, 0, 3, 0], index=list('ABCDE'))
267+
self.assertEqual(s.get('A'), 1)
268+
self.assertTrue(np.isnan(s.get('B')))
269+
self.assertEqual(s.get('C'), 0)
270+
self.assertIsNone(s.get('XX'))
271+
272+
s = pd.SparseSeries([1, np.nan, 0, 3, 0], index=list('ABCDE'),
273+
fill_value=0)
274+
self.assertEqual(s.get('A'), 1)
275+
self.assertTrue(np.isnan(s.get('B')))
276+
self.assertEqual(s.get('C'), 0)
277+
self.assertIsNone(s.get('XX'))
278+
240279
def test_take(self):
241280
orig = pd.Series([1, np.nan, np.nan, 3, np.nan],
242281
index=list('ABCDE'))
@@ -321,6 +360,53 @@ class TestSparseDataFrameIndexing(tm.TestCase):
321360

322361
_multiprocess_can_split_ = True
323362

363+
def test_getitem(self):
364+
orig = pd.DataFrame([[1, np.nan, np.nan],
365+
[2, 3, np.nan],
366+
[np.nan, np.nan, 4],
367+
[0, np.nan, 5]],
368+
columns=list('xyz'))
369+
sparse = orig.to_sparse()
370+
371+
tm.assert_sp_series_equal(sparse['x'], orig['x'].to_sparse())
372+
tm.assert_sp_frame_equal(sparse[['x']], orig[['x']].to_sparse())
373+
tm.assert_sp_frame_equal(sparse[['z', 'x']],
374+
orig[['z', 'x']].to_sparse())
375+
376+
tm.assert_sp_frame_equal(sparse[[True, False, True, True]],
377+
orig[[True, False, True, True]].to_sparse())
378+
379+
tm.assert_sp_frame_equal(sparse[[1, 2]],
380+
orig[[1, 2]].to_sparse())
381+
382+
def test_getitem_fill_value(self):
383+
orig = pd.DataFrame([[1, np.nan, 0],
384+
[2, 3, np.nan],
385+
[0, np.nan, 4],
386+
[0, np.nan, 5]],
387+
columns=list('xyz'))
388+
sparse = orig.to_sparse(fill_value=0)
389+
390+
tm.assert_sp_series_equal(sparse['y'],
391+
orig['y'].to_sparse(fill_value=0))
392+
393+
exp = orig[['x']].to_sparse(fill_value=0)
394+
exp._default_fill_value = np.nan
395+
tm.assert_sp_frame_equal(sparse[['x']], exp)
396+
397+
exp = orig[['z', 'x']].to_sparse(fill_value=0)
398+
exp._default_fill_value = np.nan
399+
tm.assert_sp_frame_equal(sparse[['z', 'x']], exp)
400+
401+
indexer = [True, False, True, True]
402+
exp = orig[indexer].to_sparse(fill_value=0)
403+
exp._default_fill_value = np.nan
404+
tm.assert_sp_frame_equal(sparse[indexer], exp)
405+
406+
exp = orig[[1, 2]].to_sparse(fill_value=0)
407+
exp._default_fill_value = np.nan
408+
tm.assert_sp_frame_equal(sparse[[1, 2]], exp)
409+
324410
def test_loc(self):
325411
orig = pd.DataFrame([[1, np.nan, np.nan],
326412
[2, 3, np.nan],
@@ -477,3 +563,151 @@ def test_iloc_slice(self):
477563
columns=list('xyz'))
478564
sparse = orig.to_sparse()
479565
tm.assert_sp_frame_equal(sparse.iloc[2:], orig.iloc[2:].to_sparse())
566+
567+
def test_at(self):
568+
orig = pd.DataFrame([[1, np.nan, 0],
569+
[2, 3, np.nan],
570+
[0, np.nan, 4],
571+
[0, np.nan, 5]],
572+
index=list('ABCD'), columns=list('xyz'))
573+
sparse = orig.to_sparse()
574+
self.assertEqual(sparse.at['A', 'x'], orig.at['A', 'x'])
575+
self.assertTrue(np.isnan(sparse.at['B', 'z']))
576+
self.assertTrue(np.isnan(sparse.at['C', 'y']))
577+
self.assertEqual(sparse.at['D', 'x'], orig.at['D', 'x'])
578+
579+
def test_at_fill_value(self):
580+
orig = pd.DataFrame([[1, np.nan, 0],
581+
[2, 3, np.nan],
582+
[0, np.nan, 4],
583+
[0, np.nan, 5]],
584+
index=list('ABCD'), columns=list('xyz'))
585+
sparse = orig.to_sparse(fill_value=0)
586+
self.assertEqual(sparse.at['A', 'x'], orig.at['A', 'x'])
587+
self.assertTrue(np.isnan(sparse.at['B', 'z']))
588+
self.assertTrue(np.isnan(sparse.at['C', 'y']))
589+
self.assertEqual(sparse.at['D', 'x'], orig.at['D', 'x'])
590+
591+
def test_iat(self):
592+
orig = pd.DataFrame([[1, np.nan, 0],
593+
[2, 3, np.nan],
594+
[0, np.nan, 4],
595+
[0, np.nan, 5]],
596+
index=list('ABCD'), columns=list('xyz'))
597+
sparse = orig.to_sparse()
598+
self.assertEqual(sparse.iat[0, 0], orig.iat[0, 0])
599+
self.assertTrue(np.isnan(sparse.iat[1, 2]))
600+
self.assertTrue(np.isnan(sparse.iat[2, 1]))
601+
self.assertEqual(sparse.iat[2, 0], orig.iat[2, 0])
602+
603+
self.assertTrue(np.isnan(sparse.iat[-1, -2]))
604+
self.assertEqual(sparse.iat[-1, -1], orig.iat[-1, -1])
605+
606+
def test_iat_fill_value(self):
607+
orig = pd.DataFrame([[1, np.nan, 0],
608+
[2, 3, np.nan],
609+
[0, np.nan, 4],
610+
[0, np.nan, 5]],
611+
index=list('ABCD'), columns=list('xyz'))
612+
sparse = orig.to_sparse(fill_value=0)
613+
self.assertEqual(sparse.iat[0, 0], orig.iat[0, 0])
614+
self.assertTrue(np.isnan(sparse.iat[1, 2]))
615+
self.assertTrue(np.isnan(sparse.iat[2, 1]))
616+
self.assertEqual(sparse.iat[2, 0], orig.iat[2, 0])
617+
618+
self.assertTrue(np.isnan(sparse.iat[-1, -2]))
619+
self.assertEqual(sparse.iat[-1, -1], orig.iat[-1, -1])
620+
621+
def test_take(self):
622+
orig = pd.DataFrame([[1, np.nan, 0],
623+
[2, 3, np.nan],
624+
[0, np.nan, 4],
625+
[0, np.nan, 5]],
626+
columns=list('xyz'))
627+
sparse = orig.to_sparse()
628+
629+
tm.assert_sp_frame_equal(sparse.take([0]),
630+
orig.take([0]).to_sparse())
631+
tm.assert_sp_frame_equal(sparse.take([0, 1]),
632+
orig.take([0, 1]).to_sparse())
633+
tm.assert_sp_frame_equal(sparse.take([-1, -2]),
634+
orig.take([-1, -2]).to_sparse())
635+
636+
def test_take_fill_value(self):
637+
orig = pd.DataFrame([[1, np.nan, 0],
638+
[2, 3, np.nan],
639+
[0, np.nan, 4],
640+
[0, np.nan, 5]],
641+
columns=list('xyz'))
642+
sparse = orig.to_sparse(fill_value=0)
643+
644+
exp = orig.take([0]).to_sparse(fill_value=0)
645+
exp._default_fill_value = np.nan
646+
tm.assert_sp_frame_equal(sparse.take([0]), exp)
647+
648+
exp = orig.take([0, 1]).to_sparse(fill_value=0)
649+
exp._default_fill_value = np.nan
650+
tm.assert_sp_frame_equal(sparse.take([0, 1]), exp)
651+
652+
exp = orig.take([-1, -2]).to_sparse(fill_value=0)
653+
exp._default_fill_value = np.nan
654+
tm.assert_sp_frame_equal(sparse.take([-1, -2]), exp)
655+
656+
def test_reindex(self):
657+
orig = pd.DataFrame([[1, np.nan, 0],
658+
[2, 3, np.nan],
659+
[0, np.nan, 4],
660+
[0, np.nan, 5]],
661+
index=list('ABCD'), columns=list('xyz'))
662+
sparse = orig.to_sparse()
663+
664+
res = sparse.reindex(['A', 'C', 'B'])
665+
exp = orig.reindex(['A', 'C', 'B']).to_sparse()
666+
tm.assert_sp_frame_equal(res, exp)
667+
668+
orig = pd.DataFrame([[np.nan, np.nan, np.nan],
669+
[np.nan, np.nan, np.nan],
670+
[np.nan, np.nan, np.nan],
671+
[np.nan, np.nan, np.nan]],
672+
index=list('ABCD'), columns=list('xyz'))
673+
sparse = orig.to_sparse()
674+
675+
res = sparse.reindex(['A', 'C', 'B'])
676+
exp = orig.reindex(['A', 'C', 'B']).to_sparse()
677+
tm.assert_sp_frame_equal(res, exp)
678+
679+
def test_reindex_fill_value(self):
680+
orig = pd.DataFrame([[1, np.nan, 0],
681+
[2, 3, np.nan],
682+
[0, np.nan, 4],
683+
[0, np.nan, 5]],
684+
index=list('ABCD'), columns=list('xyz'))
685+
sparse = orig.to_sparse(fill_value=0)
686+
687+
res = sparse.reindex(['A', 'C', 'B'])
688+
exp = orig.reindex(['A', 'C', 'B']).to_sparse(fill_value=0)
689+
tm.assert_sp_frame_equal(res, exp)
690+
691+
# all missing
692+
orig = pd.DataFrame([[np.nan, np.nan, np.nan],
693+
[np.nan, np.nan, np.nan],
694+
[np.nan, np.nan, np.nan],
695+
[np.nan, np.nan, np.nan]],
696+
index=list('ABCD'), columns=list('xyz'))
697+
sparse = orig.to_sparse(fill_value=0)
698+
699+
res = sparse.reindex(['A', 'C', 'B'])
700+
exp = orig.reindex(['A', 'C', 'B']).to_sparse(fill_value=0)
701+
tm.assert_sp_frame_equal(res, exp)
702+
703+
# all fill_value
704+
orig = pd.DataFrame([[0, 0, 0],
705+
[0, 0, 0],
706+
[0, 0, 0],
707+
[0, 0, 0]],
708+
index=list('ABCD'), columns=list('xyz'))
709+
sparse = orig.to_sparse(fill_value=0)
710+
711+
res = sparse.reindex(['A', 'C', 'B'])
712+
exp = orig.reindex(['A', 'C', 'B']).to_sparse(fill_value=0)
713+
tm.assert_sp_frame_equal(res, exp)

‎pandas/util/testing.py

+1-1
Original file line numberDiff line numberDiff line change
@@ -1263,7 +1263,7 @@ def assert_sp_frame_equal(left, right, exact_indices=True,
12631263
else:
12641264
assert_series_equal(series.to_dense(), right[col].to_dense())
12651265

1266-
assert_almost_equal(left.default_fill_value, right.default_fill_value)
1266+
assert_attr_equal('default_fill_value', left, right, obj=obj)
12671267

12681268
# do I care?
12691269
# assert(left.default_kind == right.default_kind)

0 commit comments

Comments
 (0)