def test_nominal_as_category(self):
data = pandas.DataFrame({
'attr_nominal':
pandas.Categorical(
['water', 'wine', 'beer', None, 'wine', 'water'],
ordered=False),
'attr_nominal_spaces':
pandas.Categorical([
"'red wine'", "'hard liquor'", None, 'mate', "'hard liquor'",
'mate'
],
ordered=False)
})
fp = tempfile.NamedTemporaryFile(delete=False)
try:
with TextIOWrapper(fp) as tfp:
writearff(data, tfp, relation_name='test_nominal', index=False)
with open(fp.name, 'r') as fp:
read_date = fp.readlines()
finally:
os.unlink(fp.name)
expected = EXPECTED_1.copy()
expected[
3] = "@attribute attr_nominal_spaces\t{\"hard liquor\",\"red wine\",mate}\n"
self.assertListEqual(expected, read_date)
def _make_and_write_data(fp,
n_samples,
n_features,
with_index,
with_labels,
seed,
column_prefix="V"):
x, event, time = _make_survival_data(n_samples, n_features, seed)
columns = ["{}{}".format(column_prefix, i) for i in range(n_features)]
if with_labels:
columns += ["event", "time"]
arr = numpy.column_stack((x, event, time))
else:
arr = x
if with_index:
index = numpy.arange(n_samples, dtype=numpy.float_)
numpy.random.RandomState(0).shuffle(index)
else:
index = None
dataset = pandas.DataFrame(arr, index=index, columns=columns)
dataset.index.name = "index"
writearff(dataset, fp, index=with_index)
return dataset
def test_writearff_unsupported_column_type(temp_file):
data = pandas.DataFrame({
"attr_datetime":
numpy.array([2 + 3j, 45.1 - 1j, 0 - 1j, 7 + 0j, 132 - 3j, 1 - 0.41j],
dtype="complex128")
})
with pytest.raises(TypeError, match="unsupported type complex128"):
writearff(data, temp_file, relation_name='test_delta', index=False)
def test_writearff_nominal_with_quotes(temp_file):
data = pandas.DataFrame({'attr_nominal': ['water', 'wine', 'beer', None, 'wine', 'water'],
'attr_nominal_spaces': ["'red wine'", "'hard liquor'", None, 'mate', "'hard liquor'",
'mate']})
writearff(data, temp_file, relation_name='test_nominal', index=False)
with open(temp_file.name, 'r') as fp:
read_date = fp.readlines()
assert EXPECTED_1 == read_date
def test_writearff_datetime(temp_file):
data = pandas.DataFrame(
{"attr_datetime": numpy.array(
["2014-10-31 14:13:01", "2004-03-13 19:49:31", "1998-12-06 09:10:11"], dtype="datetime64")})
writearff(data, temp_file, relation_name='test_datetime', index=False)
with open(temp_file.name, 'r') as fp:
read_date = fp.readlines()
assert EXPECTED_DATETIME == read_date
def test_writearff_nominal_as_category(temp_file):
data = pandas.DataFrame({'attr_nominal': pandas.Categorical(['water', 'wine', 'beer', None, 'wine', 'water'],
ordered=False),
'attr_nominal_spaces': pandas.Categorical(["'red wine'", "'hard liquor'", None,
'mate', "'hard liquor'", 'mate'],
ordered=False)})
writearff(data, temp_file, relation_name='test_nominal', index=False)
with open(temp_file.name, 'r') as fp:
read_date = fp.readlines()
expected = EXPECTED_1.copy()
expected[3] = "@attribute attr_nominal_spaces\t{\"hard liquor\",\"red wine\",mate}\n"
assert expected == read_date
def test_writearff_nominal_with_category_ordering(temp_file):
data = pandas.DataFrame({'attr_nominal': pandas.Categorical(['water', 'wine', 'beer', None, 'wine', 'water'],
categories=['water', 'coke', 'beer', 'wine'],
ordered=False),
'attr_nominal_spaces': ["'red wine'", "'hard liquor'", None, 'mate', "'hard liquor'",
'mate']})
writearff(data, temp_file, relation_name='test_nominal', index=False)
with open(temp_file.name, 'r') as fp:
read_date = fp.readlines()
expected = EXPECTED_1.copy()
expected[2] = "@attribute attr_nominal\t{water,coke,beer,wine}\n"
assert expected == read_date
def test_nominal_with_quotes(self):
data = pandas.DataFrame({'attr_nominal': ['water', 'wine', 'beer', None, 'wine', 'water'],
'attr_nominal_spaces': ["'red wine'", "'hard liquor'", None, 'mate', "'hard liquor'",
'mate']})
fp = tempfile.NamedTemporaryFile(delete=False)
try:
with TextIOWrapper(fp) as tfp:
writearff(data, tfp, relation_name='test_nominal', index=False)
with open(fp.name, 'r') as fp:
read_date = fp.readlines()
finally:
os.unlink(fp.name)
self.assertListEqual(EXPECTED_1, read_date)
def test_datetime(self):
data = pandas.DataFrame(
{"attr_datetime": numpy.array(
["2014-10-31 14:13:01", "2004-03-13 19:49:31", "1998-12-06 09:10:11"], dtype="datetime64")})
fp = tempfile.NamedTemporaryFile(delete=False)
try:
with TextIOWrapper(fp) as tfp:
writearff(data, tfp, relation_name='test_datetime', index=False)
with open(fp.name, 'r') as fp:
read_date = fp.readlines()
finally:
os.unlink(fp.name)
self.assertListEqual(EXPECTED_DATETIME, read_date)
