def test_fit_float(self):
"""Test the ``GreaterThan.fit`` method.
The ``GreaterThan.fit`` method should only learn and store the
``dtype`` of the ``high`` column as the ``_dtype`` attribute.
Input:
- Table that contains two constrained columns with the high one
being made of float values.
Side Effect:
- The _dtype attribute gets `float` as the value even if the low
column has a different dtype.
"""
# Setup
instance = GreaterThan(low='a', high='b')
# Run
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [4., 5., 6.],
'c': [7, 8, 9]
})
instance.fit(table_data)
# Asserts
assert instance._dtype.kind == 'f'
def test_transform_float(self):
"""Test the ``GreaterThan.transform`` method passing a high column of type float.
The ``GreaterThan.transform`` method is expected to compute the distance
between the high and low columns and replace the high column with the
logarithm of the distance + 1.
Input:
- Table with two constrained columns at a constant distance of
exactly 3 and one additional dummy column.
Output:
- Same table with the high column transformed into the logarithms
of the distances + 1, which is np.log(4).
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
# Run
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [4., 5., 6.],
'c': [7, 8, 9],
})
out = instance.transform(table_data)
# Assert
expected_out = pd.DataFrame({
'a': [1, 2, 3],
'b': [np.log(4)] * 3,
'c': [7, 8, 9],
})
pd.testing.assert_frame_equal(out, expected_out)
def test_reverse_transform(self):
"""Test the ``GreaterThan.reverse_transform`` method.
The ``GreaterThan.reverse_transform`` method is expected to:
- Return the original table data.
Input:
- Table data transformed (pandas.DataFrame)
Output:
- Table data (pandas.DataFrame)
Side effects:
- Since ``reverse_transform`` uses the class variable ``_dtype``, the ``fit`` method
must be called as well.
"""
# Setup
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [4, 5, 6],
'c': [7, 8, 9]
})
instance = GreaterThan(low='a', high='b', strict=True)
instance.fit(table_data)
# Run
out = instance.reverse_transform(table_data)
# Assert
expected_out = pd.DataFrame({
'a': [1, 2, 3],
'b': [55, 149, 405],
'c': [7, 8, 9],
})
pd.testing.assert_frame_equal(out, expected_out)
def test_transform_datetime(self):
"""Test the ``GreaterThan.transform`` method passing a high column of type datetime.
If the columns are of type datetime, ``transform`` is expected
to convert the timedelta distance into numeric before applying
the +1 and logarithm.
Input:
- Table with values at a distance of exactly 1 second.
Output:
- Same table with the high column transformed into the logarithms
of the dinstance in nanoseconds + 1, which is np.log(1_000_000_001).
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
# Run
table_data = pd.DataFrame({
'a':
pd.to_datetime(['2020-01-01T00:00:00', '2020-01-02T00:00:00']),
'b':
pd.to_datetime(['2020-01-01T00:00:01', '2020-01-02T00:00:01']),
'c': [1, 2],
})
out = instance.transform(table_data)
# Assert
expected_out = pd.DataFrame({
'a':
pd.to_datetime(['2020-01-01T00:00:00', '2020-01-02T00:00:00']),
'b': [np.log(1_000_000_001),
np.log(1_000_000_001)],
'c': [1, 2],
})
def test_reverse_transform_int(self):
"""Test the ``GreaterThan.reverse_transform`` method for dtype int.
The ``GreaterThan.reverse_transform`` method is expected to:
- apply an exponential to the input
- subtract 1
- add the low column
- convert the output to integers
Input:
- Table with a high column that contains the constant np.log(4).
Output:
- Same table with the high column replaced by the low one + 3, as int.
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
instance._dtype = pd.Series(
[1]).dtype # exact dtype (32 or 64) depends on OS
# Run
transformed = pd.DataFrame({
'a': [1, 2, 3],
'b': [np.log(4)] * 3,
'c': [7, 8, 9]
})
out = instance.reverse_transform(transformed)
# Assert
expected_out = pd.DataFrame({
'a': [1, 2, 3],
'b': [4, 5, 6],
'c': [7, 8, 9],
})
pd.testing.assert_frame_equal(out, expected_out)
def test_transform(self):
"""Test the ``GreaterThan.transform`` method.
The ``GreaterThan.transform`` method is expected to:
- Transform the original table data.
Input:
- Table data (pandas.DataFrame)
Output:
- Table data transformed (pandas.DataFrame)
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
# Run
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [4, 5, 6],
})
out = instance.transform(table_data)
# Assert
expected_out = pd.DataFrame({
'a': [1, 2, 3],
'b': [1.3862944, 1.3862944, 1.3862944]
})
pd.testing.assert_frame_equal(out, expected_out)
def test_is_valid_false_not_strict(self):
"""Test the ``GreaterThan.is_valid`` method when the column values are not valid
and the comparison is not strict.
If the columns do not satisfy the costraint, result is a series of ``False`` values.
Input:
- Table data, where the values of the ``low`` column are higher
than the values of the ``high`` column (pandas.DataFrame)
Output:
- Series of ``True`` values (pandas.Series)
"""
# Setup
instance = GreaterThan(low='a', high='b')
# Run
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [0, 1, 2],
'c': [7, 8, 9]
})
out = instance.is_valid(table_data)
# Assert
expected_out = pd.Series([False, False, False])
pd.testing.assert_series_equal(expected_out, out)
def test_reverse_transform_float(self):
"""Test the ``GreaterThan.reverse_transform`` method for dtype float.
The ``GreaterThan.reverse_transform`` method is expected to:
- apply an exponential to the input
- subtract 1
- add the low column
- convert the output to float values
Input:
- Table with a high column that contains the constant np.log(4).
Output:
- Same table with the high column replaced by the low one + 3, as float values.
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
instance._dtype = np.dtype('float')
# Run
transformed = pd.DataFrame({
'a': [1.1, 2.2, 3.3],
'b': [np.log(4)] * 3,
'c': [7, 8, 9]
})
out = instance.reverse_transform(transformed)
# Assert
expected_out = pd.DataFrame({
'a': [1.1, 2.2, 3.3],
'b': [4.1, 5.2, 6.3],
'c': [7, 8, 9],
})
pd.testing.assert_frame_equal(out, expected_out)
def test_is_valid_strict_true(self):
"""Test the ``GreaterThan.is_valid`` method with strict True.
If strict is True, equal values should count as invalid.
Input:
- Table with a strictly valid row, a strictly invalid row and
a row that has the same value for both high and low.
Output:
- True should be returned for the strictly valid row and False
for the other two.
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=False)
# Run
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [4, 2, 2],
'c': [7, 8, 9]
})
out = instance.is_valid(table_data)
# Assert
expected_out = pd.Series([True, True, False])
pd.testing.assert_series_equal(expected_out, out)
def test_reverse_transform_datetime(self):
"""Test the ``GreaterThan.reverse_transform`` method for dtype datetime.
The ``GreaterThan.reverse_transform`` method is expected to:
- apply an exponential to the input
- subtract 1
- convert the distance to a timedelta
- add the low column
- convert the output to datetimes
Input:
- Table with a high column that contains the constant np.log(1_000_000_001).
Output:
- Same table with the high column replaced by the low one + one second.
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
instance._dtype = np.dtype('<M8[ns]')
# Run
transformed = pd.DataFrame({
'a':
pd.to_datetime(['2020-01-01T00:00:00', '2020-01-02T00:00:00']),
'b': [np.log(1_000_000_001),
np.log(1_000_000_001)],
'c': [1, 2]
})
def test_transform_not_all_columns_provided(self):
"""Test the ``GreaterThan.transform`` method.
If some of the columns needed for the transform are missing, it will raise
a ``MissingConstraintColumnError``.
Input:
- Table data (pandas.DataFrame)
Output:
- Raises ``MissingConstraintColumnError``.
"""
# Setup
instance = GreaterThan(low='a', high='b', strict=True)
# Run/Assert
with pytest.raises(MissingConstraintColumnError):
instance.transform(pd.DataFrame({'a': ['a', 'b', 'c']}))
def test___init___strict_false(self):
"""Test the ``GreaterThan.__init__`` method.
The passed arguments should be stored as attributes.
Input:
- low = 'a'
- high = 'b'
Side effects:
- instance._low == 'a'
- instance._high == 'b'
- instance._strict == False
"""
# Run
instance = GreaterThan(low='a', high='b')
# Asserts
assert instance._low == 'a'
assert instance._high == 'b'
assert instance._strict is False
def test___init___strict_false(self):
"""Test the ``GreaterThan.__init__`` method.
It is expected to create a new Constraint instance and receiving ``low`` and ``high``,
names of the columns that contain the low and high value.
Input:
- low = 'a'
- high = 'b'
Side effects:
- instance._low == 'a'
- instance._high == 'b'
- instance._strict == False
"""
# Run
instance = GreaterThan(low='a', high='b')
# Asserts
assert instance._low == 'a'
assert instance._high == 'b'
assert instance._strict is False
def test___init___strict_true(self):
"""Test the ``GreaterThan.__init__`` method.
It is expected to create a new Constraint instance and receiving ``low`` and ``high``,
names of the columns that contain the low and high value. It also receives ``strict``,
a bool that indicates the comparison of the values should be strict.
Input:
- low = 'a'
- high = 'b'
- strict = True
Side effects:
- instance._low == 'a'
- instance._high == 'b'
- instance._stric == True
"""
# Run
instance = GreaterThan(low='a', high='b', strict=True)
# Asserts
assert instance._low == 'a'
assert instance._high == 'b'
assert instance._strict is True
def test_fit_datetime(self):
"""Test the ``GreaterThan.fit`` method.
The ``GreaterThan.fit`` method should only learn and store the
``dtype`` of the ``high`` column as the ``_dtype`` attribute.
Input:
- Table that contains two constrained columns of datetimes.
Side Effect:
- The _dtype attribute gets `datetime` as the value.
"""
# Setup
instance = GreaterThan(low='a', high='b')
# Run
table_data = pd.DataFrame({
'a': pd.to_datetime(['2020-01-01']),
'b': pd.to_datetime(['2020-01-02'])
})
instance.fit(table_data)
# Asserts
assert instance._dtype.kind == 'M'
def test_fit(self):
"""Test the ``GreaterThan.fit`` method.
It is expected to return the dtype of the ``high`` column.
Input:
- Table data (pandas.DataFrame)
Output:
- dtype of the ``high`` column.
"""
# Setup
instance = GreaterThan(low='a', high='b')
# Run
table_data = pd.DataFrame({
'a': [1, 2, 3],
'b': [4, 5, 6],
'c': [7, 8, 9]
})
instance.fit(table_data)
# Asserts
expected = table_data['b'].dtype
assert instance._dtype == expected
def test_conditional_sampling_constraint_uses_columns_model_reject_sampling(
column_model_mock):
"""Test that the ``sample`` method handles constraints with conditions.
The ``sample`` method is expected to properly apply constraint
transformations by sampling the missing columns for the constraint
if ``fit_columns_model`` is True. All values sampled by the column
model should be valid because reject sampling is used on any that aren't.
Setup:
- The model is being passed a ``GreaterThan`` constraint and then
asked to sample with one condition. One of the constraint columns is
the conditioned column. The ``GaussianMultivariate`` class is mocked
so that the constraint's ``_column_model`` returns some invalid rows
in order to test that the reject sampling is used.
Input:
- Conditions
Side Effects:
- Correct columns to condition on are passed to underlying sample method
"""
# Setup
constraint = GreaterThan(low='age_joined',
high='age',
handling_strategy='transform',
fit_columns_model=True,
drop='high')
data = pd.DataFrame({
'age_joined': [22.0, 21.0, 15.0, 18.0, 29.0],
'age': [27.0, 28.0, 26.0, 21.0, 30.0],
'experience_years': [6.0, 7.0, 11.0, 3.0, 7.0],
})
model = GaussianCopula(constraints=[constraint])
sampled_conditions = [
pd.DataFrame({
'age_joined': [26.0, 18.0, 31.0, 29.0, 32.0],
'age': [30.0, 30.0, 30.0, 30.0, 30.0]
}),
pd.DataFrame({
'age_joined': [28.0, 33.0, 31.0],
'age': [30.0, 30.0, 30.0]
}),
pd.DataFrame({
'age_joined': [27.0],
'age': [30.0]
})
]
column_model_mock.return_value.sample.side_effect = sampled_conditions
model.fit(data)
# Run
conditions = {'age': 30.0}
sampled_data = model.sample(5, conditions=conditions)
# Assert
assert len(column_model_mock.return_value.sample.mock_calls) == 3
expected_result = pd.DataFrame({
'age_joined': [26.0, 18.0, 29.0, 28.0, 27.0],
'age': [30.0, 30.0, 30.0, 30.0, 30.0]
})
pd.testing.assert_frame_equal(
sampled_data[['age_joined', 'age']],
expected_result[['age_joined', 'age']],
)