def test_all_missing_data(self):
mdc = NumericMetadataColumn(pd.Series(
[np.nan, np.nan, np.nan], name='col1',
index=pd.Index(['a', 'b', 'c'], name='id')))
obs = mdc.to_series()
exp = pd.Series(
[np.nan, np.nan, np.nan], name='col1',
index=pd.Index(['a', 'b', 'c'], name='id'))
pdt.assert_series_equal(obs, exp)
self.assertEqual(obs.dtype, np.float64)
def test_all_missing_data(self):
mdc = NumericMetadataColumn(
pd.Series([np.nan, np.nan, np.nan],
name='col1',
index=pd.Index(['a', 'b', 'c'], name='id')))
obs = mdc.to_series()
exp = pd.Series([np.nan, np.nan, np.nan],
name='col1',
index=pd.Index(['a', 'b', 'c'], name='id'))
pdt.assert_series_equal(obs, exp)
self.assertEqual(obs.dtype, np.float64)
def test_supported_dtype_float(self):
series = pd.Series([1.23, np.nan, 4.56, -7.891],
name='my column',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
mdc = NumericMetadataColumn(series)
self.assertEqual(mdc.id_count, 4)
self.assertEqual(mdc.id_header, 'id')
self.assertEqual(mdc.ids, ('a', 'b', 'c', 'd'))
self.assertEqual(mdc.name, 'my column')
obs_series = mdc.to_series()
pdt.assert_series_equal(obs_series, series)
self.assertEqual(obs_series.dtype, np.float64)
def test_supported_dtype_float(self):
series = pd.Series(
[1.23, np.nan, 4.56, -7.891], name='my column',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
mdc = NumericMetadataColumn(series)
self.assertEqual(mdc.id_count, 4)
self.assertEqual(mdc.id_header, 'id')
self.assertEqual(mdc.ids, ('a', 'b', 'c', 'd'))
self.assertEqual(mdc.name, 'my column')
obs_series = mdc.to_series()
pdt.assert_series_equal(obs_series, series)
self.assertEqual(obs_series.dtype, np.float64)
def test_missing_data_normalized(self):
# Different missing data representations should be normalized to np.nan
mdc = NumericMetadataColumn(pd.Series(
[np.nan, 4.2, float('nan'), -5.678], name='col1',
index=pd.Index(['a', 'b', 'c', 'd'], name='id')))
obs = mdc.to_series()
exp = pd.Series(
[np.nan, 4.2, np.nan, -5.678], name='col1',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
pdt.assert_series_equal(obs, exp)
self.assertEqual(obs.dtype, np.float64)
self.assertTrue(np.isnan(obs['a']))
self.assertTrue(np.isnan(obs['c']))
def test_unsupported_dtype(self):
with self.assertRaisesRegex(
TypeError, "NumericMetadataColumn 'col1' does not support"
".*Series.*dtype.*bool"):
NumericMetadataColumn(
pd.Series([True, False, True],
name='col1',
index=pd.Index(['a', 'b', 'c'], name='id')))
def test_missing_data_normalized(self):
# Different missing data representations should be normalized to np.nan
mdc = NumericMetadataColumn(
pd.Series([np.nan, 4.2, float('nan'), -5.678],
name='col1',
index=pd.Index(['a', 'b', 'c', 'd'], name='id')))
obs = mdc.to_series()
exp = pd.Series([np.nan, 4.2, np.nan, -5.678],
name='col1',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
pdt.assert_series_equal(obs, exp)
self.assertEqual(obs.dtype, np.float64)
self.assertTrue(np.isnan(obs['a']))
self.assertTrue(np.isnan(obs['c']))
def test_infinity_value(self):
with self.assertRaisesRegex(
ValueError, "NumericMetadataColumn.*positive or negative "
"infinity.*column 'col1'"):
NumericMetadataColumn(
pd.Series([42, float('+inf'), 4.3],
name='col1',
index=pd.Index(['a', 'b', 'c'], name='id')))
def test_numeric_metadata_column(self):
mdc = NumericMetadataColumn(
pd.Series([1e-15, 42.50, -999.0],
name='numeric-column',
index=pd.Index(['id1', 'id2', 'id3'], name='#OTU ID')))
mdc.save(self.filepath)
with open(self.filepath, 'r') as fh:
obs = fh.read()
exp = ("#OTU ID\tnumeric-column\n"
"#q2:types\tnumeric\n"
"id1\t1e-15\n"
"id2\t42.5\n"
"id3\t-999\n")
self.assertEqual(obs, exp)
def test_wrong_obj(self):
with self.assertRaisesRegex(
TypeError, 'NumericMetadataColumn constructor.*pandas.Series'):
NumericMetadataColumn(pd.DataFrame([[1, 2, 3]]))
with self.assertRaisesRegex(
TypeError,
'CategoricalMetadataColumn constructor.*pandas.Series'):
CategoricalMetadataColumn({})
def test_supported_dtype_int(self):
series = pd.Series([0, 1, 42, -2],
name='my column',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
mdc = NumericMetadataColumn(series)
self.assertEqual(mdc.id_count, 4)
self.assertEqual(mdc.id_header, 'id')
self.assertEqual(mdc.ids, ('a', 'b', 'c', 'd'))
self.assertEqual(mdc.name, 'my column')
obs_series = mdc.to_series()
exp_series = pd.Series([0.0, 1.0, 42.0, -2.0],
name='my column',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
pdt.assert_series_equal(obs_series, exp_series)
self.assertEqual(obs_series.dtype, np.float64)
def test_numeric_metadata_column(self):
mdc = NumericMetadataColumn(pd.Series(
[1e-15, 42.50, -999.0], name='numeric-column',
index=pd.Index(['id1', 'id2', 'id3'], name='#OTU ID')))
mdc.save(self.filepath)
with open(self.filepath, 'r') as fh:
obs = fh.read()
exp = (
"#OTU ID\tnumeric-column\n"
"#q2:types\tnumeric\n"
"id1\t1e-15\n"
"id2\t42.5\n"
"id3\t-999\n"
)
self.assertEqual(obs, exp)
def test_supported_dtype_int(self):
series = pd.Series(
[0, 1, 42, -2], name='my column',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
mdc = NumericMetadataColumn(series)
self.assertEqual(mdc.id_count, 4)
self.assertEqual(mdc.id_header, 'id')
self.assertEqual(mdc.ids, ('a', 'b', 'c', 'd'))
self.assertEqual(mdc.name, 'my column')
obs_series = mdc.to_series()
exp_series = pd.Series(
[0.0, 1.0, 42.0, -2.0], name='my column',
index=pd.Index(['a', 'b', 'c', 'd'], name='id'))
pdt.assert_series_equal(obs_series, exp_series)
self.assertEqual(obs_series.dtype, np.float64)
def test_type_mismatch(self):
dummy = DummyMetadataColumn(
pd.Series([1.0, 2.0, 3.0],
name='col1',
index=pd.Index(['id1', 'id2', 'id3'], name='id')))
numeric = NumericMetadataColumn(
pd.Series([1.0, 2.0, 3.0],
name='col1',
index=pd.Index(['id1', 'id2', 'id3'], name='id')))
categorical = CategoricalMetadataColumn(
pd.Series(['a', 'b', 'c'],
name='col1',
index=pd.Index(['id1', 'id2', 'id3'], name='id')))
self.assertReallyNotEqual(dummy, numeric)
self.assertReallyNotEqual(dummy, categorical)