def test_resample_data_frame_rules():
resample_rules = ['1T', '2T', '5T', '10T', '60T', '1H']
expected_lengths = [782, 392, 158, 80, 14, 14]
for resample_rule, expected_length in zip(resample_rules,
expected_lengths):
resampled_data_frame = resample_data_frame(sample_data_frame,
resample_rule)
assert len(resampled_data_frame) == expected_length
def resample(feature, data):
"""
:param feature: the feature object
:type feature: Feature
:param data: the data to be resampled
:type data: pd.DataFrame
:return: the resampled data
:rtype pd.DataFrame
"""
if not feature.local and feature.resample_minutes > 0:
resample_rule = str(feature.resample_minutes) + 'T'
sampling_function = RESAMPLE_FUNCTION_MAP.get(
feature.name, RESAMPLE_FUNCTION_LAST)
data = resample_data_frame(data,
resample_rule,
sampling_function=sampling_function)
return data
def test_resample_data_frame_functions():
index = pd.date_range('1/1/2000', periods=9, freq='T')
data_frame = pd.DataFrame(list(range(9)), index=index, columns=['col'])
resample_rule = '3T'
sampling_function = 'mean'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0., 2., 5., 7.5])
sampling_function = 'median'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0., 2., 5., 7.5])
sampling_function = 'sum'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0, 6, 15, 15])
sampling_function = 'first'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0, 1, 4, 7])
sampling_function = 'last'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0, 3, 6, 8])
sampling_function = 'min'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0, 1, 4, 7])
sampling_function = 'max'
assert_almost_equal(
resample_data_frame(data_frame, resample_rule,
sampling_function)['col'], [0, 3, 6, 8])
def test_resample_data_frame_wrong_sampling_function():
with pytest.raises(AssertionError):
resample_data_frame(sample_data_frame, '15T', 'wrong')
def test_resample_data_frame_wrong_sampling_function_type():
with pytest.raises(AssertionError):
resample_data_frame(sample_data_frame, '15T', 12)
resample_data_frame(sample_data_frame, '15T', ['mean', 'median'])
def test_resample_data_frame_wrong_resample_rule_type():
with pytest.raises(ValueError):
resample_data_frame(sample_data_frame, 15)
resample_data_frame(sample_data_frame, '15')