def test_valid(self):
sp = SparseSeries([0, 0, 0, nan, nan, 5, 6], fill_value=0)
sp_valid = sp.valid()
assert_almost_equal(sp_valid, sp.to_dense().valid())
self.assert_(sp_valid.index.equals(sp.to_dense().valid().index))
self.assertEquals(len(sp_valid.sp_values), 2)
def setUp(self):
arr, index = _test_data1()
date_index = DateRange('1/1/2011', periods=len(index))
self.bseries = SparseSeries(arr, index=index, kind='block')
self.bseries.name = 'bseries'
self.ts = self.bseries
self.btseries = SparseSeries(arr, index=date_index, kind='block')
self.iseries = SparseSeries(arr, index=index, kind='integer')
arr, index = _test_data2()
self.bseries2 = SparseSeries(arr, index=index, kind='block')
self.iseries2 = SparseSeries(arr, index=index, kind='integer')
arr, index = _test_data1_zero()
self.zbseries = SparseSeries(arr, index=index, kind='block',
fill_value=0)
self.ziseries = SparseSeries(arr, index=index, kind='integer',
fill_value=0)
arr, index = _test_data2_zero()
self.zbseries2 = SparseSeries(arr, index=index, kind='block',
fill_value=0)
self.ziseries2 = SparseSeries(arr, index=index, kind='integer',
fill_value=0)
def _check(values, index1, index2, fill_value):
first_series = SparseSeries(values, sparse_index=index1, fill_value=fill_value)
reindexed = first_series.sparse_reindex(index2)
self.assert_(reindexed.sp_index is index2)
int_indices1 = index1.to_int_index().indices
int_indices2 = index2.to_int_index().indices
expected = Series(values, index=int_indices1)
expected = expected.reindex(int_indices2).fillna(fill_value)
assert_almost_equal(expected.values, reindexed.sp_values)
def test_shift(self):
series = SparseSeries([nan, 1.0, 2.0, 3.0, nan, nan], index=np.arange(6))
shifted = series.shift(0)
self.assert_(shifted is not series)
assert_sp_series_equal(shifted, series)
f = lambda s: s.shift(1)
_dense_series_compare(series, f)
f = lambda s: s.shift(-2)
_dense_series_compare(series, f)
series = SparseSeries([nan, 1.0, 2.0, 3.0, nan, nan], index=DateRange("1/1/2000", periods=6))
f = lambda s: s.shift(2, timeRule="WEEKDAY")
_dense_series_compare(series, f)
f = lambda s: s.shift(2, offset=datetools.bday)
_dense_series_compare(series, f)
def test_take(self):
def _compare_with_dense(sp):
dense = sp.to_dense()
def _compare(idx):
dense_result = dense.take(idx).values
sparse_result = sp.take(idx)
assert_almost_equal(dense_result, sparse_result)
_compare([1.0, 2.0, 3.0, 4.0, 5.0, 0.0])
_compare([7, 2, 9, 0, 4])
_compare([3, 6, 3, 4, 7])
self._check_all(_compare_with_dense)
self.assertRaises(Exception, self.bseries.take, [-1, 0])
self.assertRaises(Exception, self.bseries.take, [0, len(self.bseries) + 1])
# Corner case
sp = SparseSeries(np.ones(10.0) * nan)
assert_almost_equal(sp.take([0, 1, 2, 3, 4]), np.repeat(nan, 5))
def setUp(self):
arr, index = _test_data1()
date_index = DateRange("1/1/2011", periods=len(index))
self.bseries = SparseSeries(arr, index=index, kind="block")
self.btseries = SparseSeries(arr, index=date_index, kind="block")
self.iseries = SparseSeries(arr, index=index, kind="integer")
arr, index = _test_data2()
self.bseries2 = SparseSeries(arr, index=index, kind="block")
self.iseries2 = SparseSeries(arr, index=index, kind="integer")
arr, index = _test_data1_zero()
self.zbseries = SparseSeries(arr, index=index, kind="block", fill_value=0)
self.ziseries = SparseSeries(arr, index=index, kind="integer", fill_value=0)
arr, index = _test_data2_zero()
self.zbseries2 = SparseSeries(arr, index=index, kind="block", fill_value=0)
self.ziseries2 = SparseSeries(arr, index=index, kind="integer", fill_value=0)
def test_constructor(self):
# test setup guys
self.assert_(np.isnan(self.bseries.fill_value))
self.assert_(isinstance(self.bseries.sp_index, BlockIndex))
self.assert_(np.isnan(self.iseries.fill_value))
self.assert_(isinstance(self.iseries.sp_index, IntIndex))
self.assertEquals(self.zbseries.fill_value, 0)
assert_equal(self.zbseries.values, self.bseries.to_dense().fillna(0))
# pass SparseSeries
s2 = SparseSeries(self.bseries)
s3 = SparseSeries(self.iseries)
s4 = SparseSeries(self.zbseries)
assert_sp_series_equal(s2, self.bseries)
assert_sp_series_equal(s3, self.iseries)
assert_sp_series_equal(s4, self.zbseries)
# Sparse time series works
date_index = DateRange('1/1/2000', periods=len(self.bseries))
s5 = SparseSeries(self.bseries, index=date_index)
self.assert_(isinstance(s5, spm.SparseTimeSeries))
# pass Series
bseries2 = SparseSeries(self.bseries.to_dense())
assert_equal(self.bseries.sp_values, bseries2.sp_values)
# pass dict?
# don't copy the data by default
values = np.ones(len(self.bseries.sp_values))
sp = SparseSeries(values, sparse_index=self.bseries.sp_index)
sp.sp_values[:5] = 97
self.assert_(values[0] == 97)
# but can make it copy!
sp = SparseSeries(values, sparse_index=self.bseries.sp_index,
copy=True)
sp.sp_values[:5] = 100
self.assert_(values[0] == 97)
class TestSparseSeries(TestCase,
test_series.CheckNameIntegration):
def setUp(self):
arr, index = _test_data1()
date_index = DateRange('1/1/2011', periods=len(index))
self.bseries = SparseSeries(arr, index=index, kind='block')
self.bseries.name = 'bseries'
self.ts = self.bseries
self.btseries = SparseSeries(arr, index=date_index, kind='block')
self.iseries = SparseSeries(arr, index=index, kind='integer')
arr, index = _test_data2()
self.bseries2 = SparseSeries(arr, index=index, kind='block')
self.iseries2 = SparseSeries(arr, index=index, kind='integer')
arr, index = _test_data1_zero()
self.zbseries = SparseSeries(arr, index=index, kind='block',
fill_value=0)
self.ziseries = SparseSeries(arr, index=index, kind='integer',
fill_value=0)
arr, index = _test_data2_zero()
self.zbseries2 = SparseSeries(arr, index=index, kind='block',
fill_value=0)
self.ziseries2 = SparseSeries(arr, index=index, kind='integer',
fill_value=0)
def test_sparse_to_dense(self):
arr, index = _test_data1()
series = self.bseries.to_dense()
assert_equal(series, arr)
series = self.bseries.to_dense(sparse_only=True)
assert_equal(series, arr[np.isfinite(arr)])
series = self.iseries.to_dense()
assert_equal(series, arr)
arr, index = _test_data1_zero()
series = self.zbseries.to_dense()
assert_equal(series, arr)
series = self.ziseries.to_dense()
assert_equal(series, arr)
def test_dense_to_sparse(self):
series = self.bseries.to_dense()
bseries = series.to_sparse(kind='block')
iseries = series.to_sparse(kind='integer')
assert_sp_series_equal(bseries, self.bseries)
assert_sp_series_equal(iseries, self.iseries)
# non-NaN fill value
series = self.zbseries.to_dense()
zbseries = series.to_sparse(kind='block', fill_value=0)
ziseries = series.to_sparse(kind='integer', fill_value=0)
assert_sp_series_equal(zbseries, self.zbseries)
assert_sp_series_equal(ziseries, self.ziseries)
def test_to_dense_preserve_name(self):
assert(self.bseries.name is not None)
result = self.bseries.to_dense()
self.assertEquals(result.name, self.bseries.name)
def test_constructor(self):
# test setup guys
self.assert_(np.isnan(self.bseries.fill_value))
self.assert_(isinstance(self.bseries.sp_index, BlockIndex))
self.assert_(np.isnan(self.iseries.fill_value))
self.assert_(isinstance(self.iseries.sp_index, IntIndex))
self.assertEquals(self.zbseries.fill_value, 0)
assert_equal(self.zbseries.values, self.bseries.to_dense().fillna(0))
# pass SparseSeries
s2 = SparseSeries(self.bseries)
s3 = SparseSeries(self.iseries)
s4 = SparseSeries(self.zbseries)
assert_sp_series_equal(s2, self.bseries)
assert_sp_series_equal(s3, self.iseries)
assert_sp_series_equal(s4, self.zbseries)
# Sparse time series works
date_index = DateRange('1/1/2000', periods=len(self.bseries))
s5 = SparseSeries(self.bseries, index=date_index)
self.assert_(isinstance(s5, spm.SparseTimeSeries))
# pass Series
bseries2 = SparseSeries(self.bseries.to_dense())
assert_equal(self.bseries.sp_values, bseries2.sp_values)
# pass dict?
# don't copy the data by default
values = np.ones(len(self.bseries.sp_values))
sp = SparseSeries(values, sparse_index=self.bseries.sp_index)
sp.sp_values[:5] = 97
self.assert_(values[0] == 97)
# but can make it copy!
sp = SparseSeries(values, sparse_index=self.bseries.sp_index,
copy=True)
sp.sp_values[:5] = 100
self.assert_(values[0] == 97)
def test_constructor_ndarray(self):
pass
def test_constructor_nonnan(self):
arr = [0, 0, 0, nan, nan]
sp_series = SparseSeries(arr, fill_value=0)
assert_equal(sp_series.values, arr)
def test_copy_astype(self):
cop = self.bseries.astype(np.float_)
self.assert_(cop is not self.bseries)
self.assert_(cop.sp_index is self.bseries.sp_index)
self.assert_(cop.dtype == np.float64)
cop2 = self.iseries.copy()
assert_sp_series_equal(cop, self.bseries)
assert_sp_series_equal(cop2, self.iseries)
# test that data is copied
cop.sp_values[:5] = 97
self.assert_(cop.sp_values[0] == 97)
self.assert_(self.bseries.sp_values[0] != 97)
# correct fill value
zbcop = self.zbseries.copy()
zicop = self.ziseries.copy()
assert_sp_series_equal(zbcop, self.zbseries)
assert_sp_series_equal(zicop, self.ziseries)
# no deep copy
view = self.bseries.copy(deep=False)
view.sp_values[:5] = 5
self.assert_((self.bseries.sp_values[:5] == 5).all())
def test_astype(self):
self.assertRaises(Exception, self.bseries.astype, np.int64)
def test_kind(self):
self.assertEquals(self.bseries.kind, 'block')
self.assertEquals(self.iseries.kind, 'integer')
def test_pickle(self):
def _test_roundtrip(series):
pickled = pickle.dumps(series, protocol=pickle.HIGHEST_PROTOCOL)
unpickled = pickle.loads(pickled)
assert_sp_series_equal(series, unpickled)
assert_series_equal(series.to_dense(), unpickled.to_dense())
self._check_all(_test_roundtrip)
def _check_all(self, check_func):
check_func(self.bseries)
check_func(self.iseries)
check_func(self.zbseries)
check_func(self.ziseries)
def test_getitem(self):
def _check_getitem(sp, dense):
for idx, val in dense.iteritems():
assert_almost_equal(val, sp[idx])
for i in xrange(len(dense)):
assert_almost_equal(sp[i], dense[i])
# j = np.float64(i)
# assert_almost_equal(sp[j], dense[j])
# negative getitem works
for i in xrange(len(dense)):
assert_almost_equal(sp[-i], dense[-i])
_check_getitem(self.bseries, self.bseries.to_dense())
_check_getitem(self.btseries, self.btseries.to_dense())
_check_getitem(self.zbseries, self.zbseries.to_dense())
_check_getitem(self.iseries, self.iseries.to_dense())
_check_getitem(self.ziseries, self.ziseries.to_dense())
# exception handling
self.assertRaises(Exception, self.bseries.__getitem__,
len(self.bseries) + 1)
# index not contained
self.assertRaises(Exception, self.btseries.__getitem__,
self.btseries.index[-1] + BDay())
def test_get(self):
assert_almost_equal(self.bseries.get(10), self.bseries[10])
self.assert_(self.bseries.get(len(self.bseries) + 1) is None)
def test_getitem_fancy_index(self):
idx = self.bseries.index
res = self.bseries[::2]
self.assert_(isinstance(res, SparseSeries))
assert_sp_series_equal(res, self.bseries.reindex(idx[::2]))
res = self.bseries[:5]
self.assert_(isinstance(res, SparseSeries))
assert_sp_series_equal(res, self.bseries.reindex(idx[:5]))
res = self.bseries[5:]
assert_sp_series_equal(res, self.bseries.reindex(idx[5:]))
def test_take(self):
def _compare_with_dense(sp):
dense = sp.to_dense()
def _compare(idx):
dense_result = dense.take(idx).values
sparse_result = sp.take(idx)
assert_almost_equal(dense_result, sparse_result)
_compare([1., 2., 3., 4., 5., 0.])
_compare([7, 2, 9, 0, 4])
_compare([3, 6, 3, 4, 7])
self._check_all(_compare_with_dense)
self.assertRaises(Exception, self.bseries.take, [-1, 0])
self.assertRaises(Exception, self.bseries.take,
[0, len(self.bseries) + 1])
# Corner case
sp = SparseSeries(np.ones(10.) * nan)
assert_almost_equal(sp.take([0, 1, 2, 3, 4]), np.repeat(nan, 5))
def test_getslice(self):
pass
def test_setitem(self):
self.assertRaises(Exception, self.bseries.__setitem__, 5, 7.)
self.assertRaises(Exception, self.iseries.__setitem__, 5, 7.)
def test_setslice(self):
self.assertRaises(Exception, self.bseries.__setslice__, 5, 10, 7.)
def test_operators(self):
def _check_op(a, b, op):
sp_result = op(a, b)
adense = a.to_dense() if isinstance(a, SparseSeries) else a
bdense = b.to_dense() if isinstance(b, SparseSeries) else b
dense_result = op(adense, bdense)
assert_almost_equal(sp_result.to_dense(), dense_result)
def check(a, b):
_check_op(a, b, operator.add)
_check_op(a, b, operator.sub)
_check_op(a, b, operator.truediv)
_check_op(a, b, operator.floordiv)
_check_op(a, b, operator.mul)
_check_op(a, b, lambda x, y: operator.add(y, x))
_check_op(a, b, lambda x, y: operator.sub(y, x))
_check_op(a, b, lambda x, y: operator.truediv(y, x))
_check_op(a, b, lambda x, y: operator.floordiv(y, x))
_check_op(a, b, lambda x, y: operator.mul(y, x))
# NaN ** 0 = 1 in C?
# _check_op(a, b, operator.pow)
# _check_op(a, b, lambda x, y: operator.pow(y, x))
check(self.bseries, self.bseries)
check(self.iseries, self.iseries)
check(self.bseries, self.iseries)
check(self.bseries, self.bseries2)
check(self.bseries, self.iseries2)
check(self.iseries, self.iseries2)
# scalar value
check(self.bseries, 5)
# zero-based
check(self.zbseries, self.zbseries * 2)
check(self.zbseries, self.zbseries2)
check(self.ziseries, self.ziseries2)
# with dense
result = self.bseries + self.bseries.to_dense()
assert_sp_series_equal(result, self.bseries + self.bseries)
# @dec.knownfailureif(True, 'Known NumPy failer as of 1.5.1')
def test_operators_corner2(self):
raise nose.SkipTest('known failer on numpy 1.5.1')
# NumPy circumvents __r*__ operations
val = np.float64(3.0)
result = val - self.zbseries
assert_sp_series_equal(result, 3 - self.zbseries)
def test_reindex(self):
def _compare_with_series(sps, new_index):
spsre = sps.reindex(new_index)
series = sps.to_dense()
seriesre = series.reindex(new_index)
seriesre = seriesre.to_sparse(fill_value=sps.fill_value)
assert_sp_series_equal(spsre, seriesre)
assert_series_equal(spsre.to_dense(), seriesre.to_dense())
_compare_with_series(self.bseries, self.bseries.index[::2])
_compare_with_series(self.bseries, list(self.bseries.index[::2]))
_compare_with_series(self.bseries, self.bseries.index[:10])
_compare_with_series(self.bseries, self.bseries.index[5:])
_compare_with_series(self.zbseries, self.zbseries.index[::2])
_compare_with_series(self.zbseries, self.zbseries.index[:10])
_compare_with_series(self.zbseries, self.zbseries.index[5:])
# special cases
same_index = self.bseries.reindex(self.bseries.index)
assert_sp_series_equal(self.bseries, same_index)
self.assert_(same_index is not self.bseries)
# corner cases
sp = SparseSeries([], index=[])
sp_zero = SparseSeries([], index=[], fill_value=0)
_compare_with_series(sp, np.arange(10))
# with copy=False
reindexed = self.bseries.reindex(self.bseries.index, copy=True)
reindexed.sp_values[:] = 1.
self.assert_((self.bseries.sp_values != 1.).all())
reindexed = self.bseries.reindex(self.bseries.index, copy=False)
reindexed.sp_values[:] = 1.
self.assert_((self.bseries.sp_values == 1.).all())
def test_sparse_reindex(self):
length = 10
def _check(values, index1, index2, fill_value):
first_series = SparseSeries(values, sparse_index=index1,
fill_value=fill_value)
reindexed = first_series.sparse_reindex(index2)
self.assert_(reindexed.sp_index is index2)
int_indices1 = index1.to_int_index().indices
int_indices2 = index2.to_int_index().indices
expected = Series(values, index=int_indices1)
expected = expected.reindex(int_indices2).fillna(fill_value)
assert_almost_equal(expected.values, reindexed.sp_values)
def _check_with_fill_value(values, first, second, fill_value=nan):
i_index1 = IntIndex(length, first)
i_index2 = IntIndex(length, second)
b_index1 = i_index1.to_block_index()
b_index2 = i_index2.to_block_index()
_check(values, i_index1, i_index2, fill_value)
_check(values, b_index1, b_index2, fill_value)
def _check_all(values, first, second):
_check_with_fill_value(values, first, second, fill_value=nan)
_check_with_fill_value(values, first, second, fill_value=0)
index1 = [2, 4, 5, 6, 8, 9]
values1 = np.arange(6.)
_check_all(values1, index1, [2, 4, 5])
_check_all(values1, index1, [2, 3, 4, 5, 6, 7, 8, 9])
_check_all(values1, index1, [0, 1])
_check_all(values1, index1, [0, 1, 7, 8, 9])
_check_all(values1, index1, [])
def test_repr(self):
bsrepr = repr(self.bseries)
isrepr = repr(self.iseries)
def test_iter(self):
pass
def test_truncate(self):
pass
def test_fillna(self):
pass
def test_groupby(self):
pass
def test_reductions(self):
def _compare_with_dense(obj, op):
sparse_result = getattr(obj, op)()
series = obj.to_dense()
dense_result = getattr(series, op)()
self.assertEquals(sparse_result, dense_result)
to_compare = ['count', 'sum', 'mean', 'std', 'var', 'skew']
def _compare_all(obj):
for op in to_compare:
_compare_with_dense(obj, op)
_compare_all(self.bseries)
self.bseries.sp_values[5:10] = np.NaN
_compare_all(self.bseries)
_compare_all(self.zbseries)
self.zbseries.sp_values[5:10] = np.NaN
_compare_all(self.zbseries)
series = self.zbseries.copy()
series.fill_value = 2
_compare_all(series)
def test_valid(self):
sp = SparseSeries([0, 0, 0, nan, nan, 5, 6],
fill_value=0)
sp_valid = sp.valid()
assert_almost_equal(sp_valid, sp.to_dense().valid())
self.assert_(sp_valid.index.equals(sp.to_dense().valid().index))
self.assertEquals(len(sp_valid.sp_values), 2)
def test_homogenize(self):
def _check_matches(indices, expected):
data = {}
for i, idx in enumerate(indices):
data[i] = SparseSeries(idx.to_int_index().indices,
sparse_index=idx)
homogenized = spm.homogenize(data)
for k, v in homogenized.iteritems():
assert(v.sp_index.equals(expected))
indices1 = [BlockIndex(10, [2], [7]),
BlockIndex(10, [1, 6], [3, 4]),
BlockIndex(10, [0], [10])]
expected1 = BlockIndex(10, [2, 6], [2, 3])
_check_matches(indices1, expected1)
indices2 = [BlockIndex(10, [2], [7]),
BlockIndex(10, [2], [7])]
expected2 = indices2[0]
_check_matches(indices2, expected2)
# must have NaN fill value
data = {'a' : SparseSeries(np.arange(7), sparse_index=expected2,
fill_value=0)}
nose.tools.assert_raises(Exception, spm.homogenize, data)
def test_fill_value_corner(self):
cop = self.zbseries.copy()
cop.fill_value = 0
result = self.bseries / cop
self.assert_(np.isnan(result.fill_value))
cop2 = self.zbseries.copy()
cop2.fill_value = 1
result = cop2 / cop
self.assert_(np.isnan(result.fill_value))
def test_shift(self):
series = SparseSeries([nan, 1., 2., 3., nan, nan],
index=np.arange(6))
shifted = series.shift(0)
self.assert_(shifted is not series)
assert_sp_series_equal(shifted, series)
f = lambda s: s.shift(1)
_dense_series_compare(series, f)
f = lambda s: s.shift(-2)
_dense_series_compare(series, f)
series = SparseSeries([nan, 1., 2., 3., nan, nan],
index=DateRange('1/1/2000', periods=6))
f = lambda s: s.shift(2, timeRule='WEEKDAY')
_dense_series_compare(series, f)
f = lambda s: s.shift(2, offset=datetools.bday)
_dense_series_compare(series, f)
def test_cumsum(self):
result = self.bseries.cumsum()
expected = self.bseries.to_dense().cumsum()
self.assert_(isinstance(result, SparseSeries))
assert_series_equal(result.to_dense(), expected)
result = self.zbseries.cumsum()
expected = self.zbseries.to_dense().cumsum()
self.assert_(isinstance(result, Series))
assert_series_equal(result, expected)
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 _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)
