def setUp(self):
self.data = {'A' : [nan, nan, nan, 0, 1, 2, 3, 4, 5, 6],
'B' : [0, 1, 2, nan, nan, nan, 3, 4, 5, 6],
'C' : np.arange(10),
'D' : [0, 1, 2, 3, 4, 5, nan, nan, nan, nan]}
self.dates = DateRange('1/1/2011', periods=10)
self.frame = SparseDataFrame(self.data, index=self.dates)
self.iframe = SparseDataFrame(self.data, index=self.dates,
default_kind='integer')
values = self.frame.values.copy()
values[np.isnan(values)] = 0
self.zframe = SparseDataFrame(values, columns=['A', 'B', 'C', 'D'],
default_fill_value=0,
index=self.dates)
values = self.frame.values.copy()
values[np.isnan(values)] = 2
self.fill_frame = SparseDataFrame(values, columns=['A', 'B', 'C', 'D'],
default_fill_value=2,
index=self.dates)
self.empty = SparseDataFrame()
def test_as_matrix(self):
empty = self.empty.as_matrix()
self.assert_(empty.shape == (0, 0))
no_cols = SparseDataFrame(index=np.arange(10))
mat = no_cols.as_matrix()
self.assert_(mat.shape == (10, 0))
no_index = SparseDataFrame(columns=np.arange(10))
mat = no_index.as_matrix()
self.assert_(mat.shape == (0, 10))
def get_empty_Frame(data, sparse):
if isinstance(data, Series):
index = data.index
else:
index = np.arange(len(data))
if not sparse:
return DataFrame(index=index)
else:
return SparseDataFrame(index=index)
def setUp(self):
self.data = {
"A": [nan, nan, nan, 0, 1, 2, 3, 4, 5, 6],
"B": [0, 1, 2, nan, nan, nan, 3, 4, 5, 6],
"C": np.arange(10),
"D": [0, 1, 2, 3, 4, 5, nan, nan, nan, nan],
}
self.dates = DateRange("1/1/2011", periods=10)
self.frame = SparseDataFrame(self.data, index=self.dates)
self.iframe = SparseDataFrame(self.data, index=self.dates, default_kind="integer")
values = self.frame.values.copy()
values[np.isnan(values)] = 0
self.zframe = SparseDataFrame(values, columns=["A", "B", "C", "D"], default_fill_value=0, index=self.dates)
values = self.frame.values.copy()
values[np.isnan(values)] = 2
self.fill_frame = SparseDataFrame(values, columns=["A", "B", "C", "D"], default_fill_value=2, index=self.dates)
self.empty = SparseDataFrame()
def _get_dummies_1d(data,
prefix,
prefix_sep='_',
dummy_na=False,
sparse=False,
drop_first=False):
# Series avoids inconsistent NaN handling
codes, levels = _factorize_from_iterable(Series(data))
def get_empty_Frame(data, sparse):
if isinstance(data, Series):
index = data.index
else:
index = np.arange(len(data))
if not sparse:
return DataFrame(index=index)
else:
return SparseDataFrame(index=index)
# if all NaN
if not dummy_na and len(levels) == 0:
return get_empty_Frame(data, sparse)
codes = codes.copy()
if dummy_na:
codes[codes == -1] = len(levels)
levels = np.append(levels, np.nan)
# if dummy_na, we just fake a nan level. drop_first will drop it again
if drop_first and len(levels) == 1:
return get_empty_Frame(data, sparse)
number_of_cols = len(levels)
if prefix is not None:
dummy_cols = ['%s%s%s' % (prefix, prefix_sep, v) for v in levels]
else:
dummy_cols = levels
if isinstance(data, Series):
index = data.index
else:
index = None
if sparse:
sparse_series = {}
N = len(data)
sp_indices = [[] for _ in range(len(dummy_cols))]
for ndx, code in enumerate(codes):
if code == -1:
# Blank entries if not dummy_na and code == -1, #GH4446
continue
sp_indices[code].append(ndx)
if drop_first:
# remove first categorical level to avoid perfect collinearity
# GH12042
sp_indices = sp_indices[1:]
dummy_cols = dummy_cols[1:]
for col, ixs in zip(dummy_cols, sp_indices):
sarr = SparseArray(np.ones(len(ixs), dtype=np.uint8),
sparse_index=IntIndex(N, ixs),
fill_value=0,
dtype=np.uint8)
sparse_series[col] = SparseSeries(data=sarr, index=index)
out = SparseDataFrame(sparse_series,
index=index,
columns=dummy_cols,
dtype=np.uint8)
return out
else:
dummy_mat = np.eye(number_of_cols, dtype=np.uint8).take(codes, axis=0)
if not dummy_na:
# reset NaN GH4446
dummy_mat[codes == -1] = 0
if drop_first:
# remove first GH12042
dummy_mat = dummy_mat[:, 1:]
dummy_cols = dummy_cols[1:]
return DataFrame(dummy_mat, index=index, columns=dummy_cols)
def _get_dummies_1d(data,
prefix,
prefix_sep='_',
dummy_na=False,
sparse=False):
# Series avoids inconsistent NaN handling
cat = Categorical.from_array(Series(data), ordered=True)
levels = cat.categories
# if all NaN
if not dummy_na and len(levels) == 0:
if isinstance(data, Series):
index = data.index
else:
index = np.arange(len(data))
if not sparse:
return DataFrame(index=index)
else:
return SparseDataFrame(index=index)
codes = cat.codes.copy()
if dummy_na:
codes[codes == -1] = len(cat.categories)
levels = np.append(cat.categories, np.nan)
number_of_cols = len(levels)
if prefix is not None:
dummy_cols = ['%s%s%s' % (prefix, prefix_sep, v) for v in levels]
else:
dummy_cols = levels
if isinstance(data, Series):
index = data.index
else:
index = None
if sparse:
sparse_series = {}
N = len(data)
sp_indices = [[] for _ in range(len(dummy_cols))]
for ndx, code in enumerate(codes):
if code == -1:
# Blank entries if not dummy_na and code == -1, #GH4446
continue
sp_indices[code].append(ndx)
for col, ixs in zip(dummy_cols, sp_indices):
sarr = SparseArray(np.ones(len(ixs)),
sparse_index=IntIndex(N, ixs),
fill_value=0)
sparse_series[col] = SparseSeries(data=sarr, index=index)
return SparseDataFrame(sparse_series, index=index, columns=dummy_cols)
else:
dummy_mat = np.eye(number_of_cols).take(codes, axis=0)
if not dummy_na:
# reset NaN GH4446
dummy_mat[codes == -1] = 0
return DataFrame(dummy_mat, index=index, columns=dummy_cols)
def time_sparse_frame_constructor(self):
SparseDataFrame(columns=np.arange(100), index=np.arange(1000))
def time_sparse_series_to_frame(self):
SparseDataFrame(self.series)
class TestSparseDataFrame(TestCase, test_frame.SafeForSparse):
klass = SparseDataFrame
def setUp(self):
self.data = {'A' : [nan, nan, nan, 0, 1, 2, 3, 4, 5, 6],
'B' : [0, 1, 2, nan, nan, nan, 3, 4, 5, 6],
'C' : np.arange(10),
'D' : [0, 1, 2, 3, 4, 5, nan, nan, nan, nan]}
self.dates = DateRange('1/1/2011', periods=10)
self.frame = SparseDataFrame(self.data, index=self.dates)
self.iframe = SparseDataFrame(self.data, index=self.dates,
default_kind='integer')
values = self.frame.values.copy()
values[np.isnan(values)] = 0
self.zframe = SparseDataFrame(values, columns=['A', 'B', 'C', 'D'],
default_fill_value=0,
index=self.dates)
values = self.frame.values.copy()
values[np.isnan(values)] = 2
self.fill_frame = SparseDataFrame(values, columns=['A', 'B', 'C', 'D'],
default_fill_value=2,
index=self.dates)
self.empty = SparseDataFrame()
def test_as_matrix(self):
empty = self.empty.as_matrix()
self.assert_(empty.shape == (0, 0))
no_cols = SparseDataFrame(index=np.arange(10))
mat = no_cols.as_matrix()
self.assert_(mat.shape == (10, 0))
no_index = SparseDataFrame(columns=np.arange(10))
mat = no_index.as_matrix()
self.assert_(mat.shape == (0, 10))
def test_copy(self):
cp = self.frame.copy()
self.assert_(isinstance(cp, SparseDataFrame))
assert_sp_frame_equal(cp, self.frame)
self.assert_(cp.index is self.frame.index)
def test_constructor(self):
for col, series in self.frame.iteritems():
self.assert_(isinstance(series, SparseSeries))
self.assert_(isinstance(self.iframe['A'].sp_index, IntIndex))
# constructed zframe from matrix above
self.assertEquals(self.zframe['A'].fill_value, 0)
assert_almost_equal([0, 0, 0, 0, 1, 2, 3, 4, 5, 6],
self.zframe['A'].values)
# construct from nested dict
data = {}
for c, s in self.frame.iteritems():
data[c] = s.to_dict()
sdf = SparseDataFrame(data)
assert_sp_frame_equal(sdf, self.frame)
# TODO: test data is copied from inputs
# init dict with different index
idx = self.frame.index[:5]
cons = SparseDataFrame(self.frame._series, index=idx,
columns=self.frame.columns,
default_fill_value=self.frame.default_fill_value,
default_kind=self.frame.default_kind)
reindexed = self.frame.reindex(idx)
assert_sp_frame_equal(cons, reindexed)
def test_constructor_ndarray(self):
# no index or columns
sp = SparseDataFrame(self.frame.values)
# 1d
sp = SparseDataFrame(self.data['A'], index=self.dates,
columns=['A'])
assert_sp_frame_equal(sp, self.frame.reindex(columns=['A']))
# wrong length index / columns
self.assertRaises(Exception, SparseDataFrame, self.frame.values,
index=self.frame.index[:-1])
self.assertRaises(Exception, SparseDataFrame, self.frame.values,
columns=self.frame.columns[:-1])
def test_constructor_empty(self):
sp = SparseDataFrame()
self.assert_(len(sp.index) == 0)
self.assert_(len(sp.columns) == 0)
def test_constructor_dataframe(self):
dense = self.frame.to_dense()
sp = SparseDataFrame(dense)
assert_sp_frame_equal(sp, self.frame)
def test_array_interface(self):
res = np.sqrt(self.frame)
dres = np.sqrt(self.frame.to_dense())
assert_frame_equal(res.to_dense(), dres)
def test_pickle(self):
def _test_roundtrip(frame):
pickled = pickle.dumps(frame, protocol=pickle.HIGHEST_PROTOCOL)
unpickled = pickle.loads(pickled)
assert_sp_frame_equal(frame, unpickled)
self._check_all(_test_roundtrip)
def test_dense_to_sparse(self):
df = DataFrame({'A' : [nan, nan, nan, 1, 2],
'B' : [1, 2, nan, nan, nan]})
sdf = df.to_sparse()
self.assert_(isinstance(sdf, SparseDataFrame))
self.assert_(np.isnan(sdf.default_fill_value))
self.assert_(isinstance(sdf['A'].sp_index, BlockIndex))
testing.assert_frame_equal(sdf.to_dense(), df)
sdf = df.to_sparse(kind='integer')
self.assert_(isinstance(sdf['A'].sp_index, IntIndex))
df = DataFrame({'A' : [0, 0, 0, 1, 2],
'B' : [1, 2, 0, 0, 0]}, dtype=float)
sdf = df.to_sparse(fill_value=0)
self.assertEquals(sdf.default_fill_value, 0)
testing.assert_frame_equal(sdf.to_dense(), df)
def test_sparse_to_dense(self):
pass
def test_sparse_series_ops(self):
self._check_all(self._check_frame_ops)
def _check_frame_ops(self, frame):
fill = frame.default_fill_value
def _compare_to_dense(a, b, da, db, op):
sparse_result = op(a, b)
dense_result = op(da, db)
dense_result = dense_result.to_sparse(fill_value=fill)
assert_sp_frame_equal(sparse_result, dense_result,
exact_indices=False)
if isinstance(a, DataFrame) and isinstance(db, DataFrame):
mixed_result = op(a, db)
self.assert_(isinstance(mixed_result, SparseDataFrame))
assert_sp_frame_equal(mixed_result, sparse_result,
exact_indices=False)
opnames = ['add', 'sub', 'mul', 'truediv', 'floordiv']
ops = [getattr(operator, name) for name in opnames]
fidx = frame.index
# time series operations
series = [frame['A'], frame['B'],
frame['C'], frame['D'],
frame['A'].reindex(fidx[:7]),
frame['A'].reindex(fidx[::2]),
SparseSeries([], index=[])]
for op in ops:
_compare_to_dense(frame, frame[::2], frame.to_dense(),
frame[::2].to_dense(), op)
for s in series:
_compare_to_dense(frame, s, frame.to_dense(),
s.to_dense(), op)
_compare_to_dense(s, frame, s.to_dense(),
frame.to_dense(), op)
# cross-sectional operations
series = [frame.xs(fidx[0]),
frame.xs(fidx[3]),
frame.xs(fidx[5]),
frame.xs(fidx[7]),
frame.xs(fidx[5])[:2]]
for op in ops:
for s in series:
_compare_to_dense(frame, s, frame.to_dense(),
s, op)
_compare_to_dense(s, frame, s,
frame.to_dense(), op)
def test_op_corners(self):
empty = self.empty + self.empty
self.assert_(not empty)
foo = self.frame + self.empty
assert_sp_frame_equal(foo, self.frame * np.nan)
foo = self.empty + self.frame
assert_sp_frame_equal(foo, self.frame * np.nan)
def test_scalar_ops(self):
pass
def test_getitem(self):
pass
def test_fancy_index_misc(self):
# axis = 0
sliced = self.frame.ix[-2:, :]
expected = self.frame.reindex(index=self.frame.index[-2:])
assert_sp_frame_equal(sliced, expected)
# axis = 1
sliced = self.frame.ix[:, -2:]
expected = self.frame.reindex(columns=self.frame.columns[-2:])
assert_sp_frame_equal(sliced, expected)
def test_getitem_overload(self):
# slicing
sl = self.frame[:20]
assert_sp_frame_equal(sl, self.frame.reindex(self.frame.index[:20]))
# boolean indexing
d = self.frame.index[5]
indexer = self.frame.index > d
subindex = self.frame.index[indexer]
subframe = self.frame[indexer]
self.assert_(np.array_equal(subindex, subframe.index))
self.assertRaises(Exception, self.frame.__getitem__, indexer[:-1])
def test_setitem(self):
def _check_frame(frame):
N = len(frame)
# insert SparseSeries
frame['E'] = frame['A']
self.assert_(isinstance(frame['E'], SparseSeries))
assert_sp_series_equal(frame['E'], frame['A'])
# insert SparseSeries differently-indexed
to_insert = frame['A'][::2]
frame['E'] = to_insert
assert_series_equal(frame['E'].to_dense(),
to_insert.to_dense().reindex(frame.index))
# insert Series
frame['F'] = frame['A'].to_dense()
self.assert_(isinstance(frame['F'], SparseSeries))
assert_sp_series_equal(frame['F'], frame['A'])
# insert Series differently-indexed
to_insert = frame['A'].to_dense()[::2]
frame['G'] = to_insert
assert_series_equal(frame['G'].to_dense(),
to_insert.reindex(frame.index))
# insert ndarray
frame['H'] = np.random.randn(N)
self.assert_(isinstance(frame['H'], SparseSeries))
to_sparsify = np.random.randn(N)
to_sparsify[N // 2:] = frame.default_fill_value
frame['I'] = to_sparsify
self.assertEquals(len(frame['I'].sp_values), N // 2)
# insert ndarray wrong size
self.assertRaises(Exception, frame.__setitem__, 'foo',
np.random.randn(N - 1))
# scalar value
frame['J'] = 5
self.assertEquals(len(frame['J'].sp_values), N)
self.assert_((frame['J'].sp_values == 5).all())
frame['K'] = frame.default_fill_value
self.assertEquals(len(frame['K'].sp_values), 0)
self._check_all(_check_frame)
def test_setitem_corner(self):
self.frame['a'] = self.frame['B']
assert_sp_series_equal(self.frame['a'], self.frame['B'])
def test_delitem(self):
A = self.frame['A']
C = self.frame['C']
del self.frame['B']
self.assert_('B' not in self.frame)
assert_sp_series_equal(self.frame['A'], A)
assert_sp_series_equal(self.frame['C'], C)
del self.frame['D']
self.assert_('D' not in self.frame)
del self.frame['A']
self.assert_('A' not in self.frame)
def test_set_columns(self):
self.frame.columns = self.frame.columns
self.assertRaises(Exception, setattr, self.frame, 'columns',
self.frame.columns[:-1])
def test_set_index(self):
self.frame.index = self.frame.index
self.assertRaises(Exception, setattr, self.frame, 'index',
self.frame.index[:-1])
def test_append(self):
a = self.frame[:5]
b = self.frame[5:]
appended = a.append(b)
assert_sp_frame_equal(appended, self.frame)
a = self.frame.ix[:5, :3]
b = self.frame.ix[5:]
appended = a.append(b)
assert_sp_frame_equal(appended.ix[:, :3], self.frame.ix[:, :3])
def test_apply(self):
applied = self.frame.apply(np.sqrt)
self.assert_(isinstance(applied, SparseDataFrame))
assert_almost_equal(applied.values, np.sqrt(self.frame.values))
applied = self.fill_frame.apply(np.sqrt)
self.assert_(applied['A'].fill_value == np.sqrt(2))
# agg / broadcast
applied = self.frame.apply(np.sum)
assert_series_equal(applied,
self.frame.to_dense().apply(np.sum))
broadcasted = self.frame.apply(np.sum, broadcast=True)
self.assert_(isinstance(broadcasted, SparseDataFrame))
assert_frame_equal(broadcasted.to_dense(),
self.frame.to_dense().apply(np.sum, broadcast=True))
self.assert_(self.empty.apply(np.sqrt) is self.empty)
def test_applymap(self):
# just test that it works
result = self.frame.applymap(lambda x: x * 2)
self.assert_(isinstance(result, SparseDataFrame))
def test_astype(self):
self.assertRaises(Exception, self.frame.astype, np.int64)
def test_fillna(self):
self.assertRaises(NotImplementedError, self.frame.fillna, 0)
def test_rename(self):
# just check this works
renamed = self.frame.rename(index=str)
renamed = self.frame.rename(columns=lambda x: '%s%d' % (x, len(x)))
def test_corr(self):
res = self.frame.corr()
assert_frame_equal(res, self.frame.to_dense().corr())
def test_describe(self):
self.frame['foo'] = np.nan
desc = self.frame.describe()
def test_join(self):
left = self.frame.ix[:, ['A', 'B']]
right = self.frame.ix[:, ['C', 'D']]
joined = left.join(right)
assert_sp_frame_equal(joined, self.frame)
right = self.frame.ix[:, ['B', 'D']]
self.assertRaises(Exception, left.join, right)
def test_reindex(self):
def _check_frame(frame):
index = frame.index
sidx = index[::2]
sidx2 = index[:5]
sparse_result = frame.reindex(sidx)
dense_result = frame.to_dense().reindex(sidx)
assert_frame_equal(sparse_result.to_dense(), dense_result)
assert_frame_equal(frame.reindex(list(sidx)).to_dense(),
dense_result)
sparse_result2 = sparse_result.reindex(index)
dense_result2 = dense_result.reindex(index)
assert_frame_equal(sparse_result2.to_dense(), dense_result2)
# propagate CORRECT fill value
assert_almost_equal(sparse_result.default_fill_value,
frame.default_fill_value)
assert_almost_equal(sparse_result['A'].fill_value,
frame['A'].fill_value)
# length zero
length_zero = frame.reindex([])
self.assertEquals(len(length_zero), 0)
self.assertEquals(len(length_zero.columns), len(frame.columns))
self.assertEquals(len(length_zero['A']), 0)
# frame being reindexed has length zero
length_n = length_zero.reindex(index)
self.assertEquals(len(length_n), len(frame))
self.assertEquals(len(length_n.columns), len(frame.columns))
self.assertEquals(len(length_n['A']), len(frame))
# reindex columns
reindexed = frame.reindex(columns=['A', 'B', 'Z'])
self.assertEquals(len(reindexed.columns), 3)
assert_almost_equal(reindexed['Z'].fill_value,
frame.default_fill_value)
self.assert_(np.isnan(reindexed['Z'].sp_values).all())
_check_frame(self.frame)
_check_frame(self.iframe)
_check_frame(self.zframe)
_check_frame(self.fill_frame)
# with copy=False
reindexed = self.frame.reindex(self.frame.index, copy=False)
reindexed['F'] = reindexed['A']
self.assert_('F' in self.frame)
reindexed = self.frame.reindex(self.frame.index)
reindexed['G'] = reindexed['A']
self.assert_('G' not in self.frame)
def test_density(self):
df = SparseDataFrame({'A' : [nan, nan, nan, 0, 1, 2, 3, 4, 5, 6],
'B' : [0, 1, 2, nan, nan, nan, 3, 4, 5, 6],
'C' : np.arange(10),
'D' : [0, 1, 2, 3, 4, 5, nan, nan, nan, nan]})
self.assertEquals(df.density, 0.75)
def test_to_dense(self):
def _check(frame):
dense_dm = frame.to_dense()
assert_frame_equal(frame, dense_dm)
self._check_all(_check)
def test_stack_sparse_frame(self):
def _check(frame):
dense_frame = frame.to_dense()
wp = Panel.from_dict({'foo' : frame})
from_dense_lp = wp.to_long()
from_sparse_lp = spm.stack_sparse_frame(frame)
self.assert_(np.array_equal(from_dense_lp.values,
from_sparse_lp.values))
_check(self.frame)
_check(self.iframe)
# for now
self.assertRaises(Exception, _check, self.zframe)
self.assertRaises(Exception, _check, self.fill_frame)
def test_transpose(self):
def _check(frame):
transposed = frame.T
untransposed = transposed.T
assert_sp_frame_equal(frame, untransposed)
self._check_all(_check)
def test_shift(self):
def _check(frame):
shifted = frame.shift(0)
self.assert_(shifted is not frame)
assert_sp_frame_equal(shifted, frame)
f = lambda s: s.shift(1)
_dense_frame_compare(frame, f)
f = lambda s: s.shift(-2)
_dense_frame_compare(frame, f)
f = lambda s: s.shift(2, timeRule='WEEKDAY')
_dense_frame_compare(frame, f)
f = lambda s: s.shift(2, offset=datetools.bday)
_dense_frame_compare(frame, f)
self._check_all(_check)
def test_count(self):
result = self.frame.count()
dense_result = self.frame.to_dense().count()
assert_series_equal(result, dense_result)
result = self.frame.count(1)
dense_result = self.frame.to_dense().count(1)
assert_series_equal(result, dense_result)
def test_cumsum(self):
result = self.frame.cumsum()
expected = self.frame.to_dense().cumsum()
self.assert_(isinstance(result, SparseDataFrame))
assert_frame_equal(result.to_dense(), expected)
def _check_all(self, check_func):
check_func(self.frame)
check_func(self.iframe)
check_func(self.zframe)
check_func(self.fill_frame)