def test_get_tfidf(fn):
def sort_ndarray(array):
hashes = [hash(str(x)) for x in array]
sotred_indicies = sorted(range(len(hashes)), key=lambda k: hashes[k])
return array[sotred_indicies]
fn_inputs = {
"sentiment_words":
pd.Series(["one", "last", "second"]),
"docs": [
"this is a document",
"this document is the second document",
"last one",
],
}
fn_correct_outputs = OrderedDict([(
"tfidf",
np.array([[0.0, 0.0, 0.0], [1.0, 0.0, 0.0],
[0.0, 0.70710678, 0.70710678]]),
)])
fn_out = fn(**fn_inputs)
assert_structure(fn_out, fn_correct_outputs["tfidf"], "tfidf")
assert np.isclose(sort_ndarray(fn_out.T),
sort_ndarray(fn_correct_outputs["tfidf"].T)).all(), (
"Wrong value for tfidf.\n"
"INPUT docs:\n{}\n\n"
"OUTPUT tfidf:\n{}\n\n"
"A POSSIBLE CORRECT OUTPUT FOR tfidf:\n{}\n".format(
fn_inputs["docs"], fn_out,
fn_correct_outputs["tfidf"]))
def test_get_bag_of_words(fn):
def sort_ndarray(array):
hashes = [hash(str(x)) for x in array]
sotred_indicies = sorted(range(len(hashes)), key=lambda k: hashes[k])
return array[sotred_indicies]
fn_inputs = {
'docs': [
'this is a document', 'this document is the second document',
'last one'
]
}
fn_correct_outputs = OrderedDict([('bag_of_words',
np.array([[0, 1, 1, 0, 0, 0, 1],
[0, 1, 2, 0, 1, 1, 1],
[1, 0, 0, 1, 0, 0, 0]]))])
fn_out = fn(**fn_inputs)
assert_structure(fn_out, fn_correct_outputs['bag_of_words'],
'bag_of_words')
assert np.array_equal(sort_ndarray(fn_out.T), sort_ndarray(fn_correct_outputs['bag_of_words'].T)), \
'Wrong value for bag_of_words.\n' \
'INPUT docs:\n{}\n\n' \
'OUTPUT bag_of_words:\n{}\n\n' \
'A POSSIBLE CORRECT OUTPUT FOR bag_of_words:\n{}\n'\
.format(fn_inputs['docs'], fn_out, fn_correct_outputs['bag_of_words'])
def test_get_bag_of_words(fn):
def sort_ndarray(array):
hashes = [hash(str(x)) for x in array]
sotred_indicies = sorted(range(len(hashes)), key=lambda k: hashes[k])
return array[sotred_indicies]
fn_inputs = {
"sentiment_words":
pd.Series(["one", "last", "second"]),
"docs": [
"this is a document",
"this document is the second document",
"last one",
],
}
fn_correct_outputs = OrderedDict([("bag_of_words",
np.array([[0, 0, 0], [1, 0, 0],
[0, 1, 1]]))])
fn_out = fn(**fn_inputs)
assert_structure(fn_out, fn_correct_outputs["bag_of_words"],
"bag_of_words")
assert np.array_equal(sort_ndarray(
fn_out.T), sort_ndarray(fn_correct_outputs["bag_of_words"].T)), (
"Wrong value for bag_of_words.\n"
"INPUT docs:\n{}\n\n"
"OUTPUT bag_of_words:\n{}\n\n"
"A POSSIBLE CORRECT OUTPUT FOR bag_of_words:\n{}\n".format(
fn_inputs["docs"], fn_out, fn_correct_outputs["bag_of_words"]))
def test_get_tfidf(fn):
def sort_ndarray(array):
hashes = [hash(str(x)) for x in array]
sotred_indicies = sorted(range(len(hashes)), key=lambda k: hashes[k])
return array[sotred_indicies]
fn_inputs = {
'sentiment_words':
pd.Series(['one', 'last', 'second']),
'docs': [
'this is a document', 'this document is the second document',
'last one'
]
}
fn_correct_outputs = OrderedDict([('tfidf',
np.array([[0.0, 0.0, 0.0],
[1.0, 0.0, 0.0],
[0.0, 0.70710678,
0.70710678]]))])
fn_out = fn(**fn_inputs)
assert_structure(fn_out, fn_correct_outputs['tfidf'], 'tfidf')
assert np.isclose(sort_ndarray(fn_out.T), sort_ndarray(fn_correct_outputs['tfidf'].T)).all(), \
'Wrong value for tfidf.\n' \
'INPUT docs:\n{}\n\n' \
'OUTPUT tfidf:\n{}\n\n' \
'A POSSIBLE CORRECT OUTPUT FOR tfidf:\n{}\n'\
.format(fn_inputs['docs'], fn_out, fn_correct_outputs['tfidf'])
def test_non_overlapping_estimators(fn):
n_estimators = 3
columns = ['test column 1', 'test column 2']
dates = generate_random_dates(8)
assets = get_assets(3)
index = pd.MultiIndex.from_product([dates, assets])
noise = np.random.RandomState(0).random_sample([len(index)]) * len(index)
values = np.arange(len(index) * len(columns)).reshape([len(columns), len(index)]).T
targets = np.sum(values, axis=-1) + noise
classifiers = [
RandomForestRegressor(300, oob_score=True, n_jobs=-1, random_state=101)
for _ in range(n_estimators)]
fn_inputs = {
'x': pd.DataFrame(values, index, columns),
'y': pd.Series(targets, index),
'classifiers': classifiers,
'n_skip_samples': 3}
random_forest_regressor_fit = RandomForestRegressor.fit
with patch.object(RandomForestRegressor, 'fit', autospec=True) as mock_fit:
mock_fit.side_effect = random_forest_regressor_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, [RandomForestRegressor for _ in range(n_estimators)], 'PCA')
for classifier in fn_return_value:
try:
classifier.fit.assert_called()
except AssertionError:
raise Exception('Test Failure: RandomForestRegressor.fit not called on all classifiers')
def test_fit_pca(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns':
pd.DataFrame([[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734]], dates, assets),
'num_factor_exposures':
2,
'svd_solver':
'full'
}
fn_correct_values = {
'PCA':
PCA(),
'PCA.components_':
np.array([[0.81925896, -0.40427891, 0.40666118],
[-0.02011128, 0.68848693, 0.72496985]])
}
pca_fit = PCA.fit
with patch.object(PCA, 'fit', autospec=True) as mock_fit:
mock_fit.side_effect = pca_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, fn_correct_values['PCA'], 'PCA')
try:
# print(dir(fn_return_value.fit))
fn_return_value.fit.assert_called()
# old python:
# fn_return_value.fit.assert_any_call()
except AssertionError:
raise Exception('Test Failure: PCA.fit not called')
try:
fn_return_value.fit.assert_called_with(self=fn_return_value,
X=fn_inputs['returns'])
except Exception:
raise Exception(
'Test Failure: PCA.fit called with the wrong arguments')
assert_structure(fn_return_value.components_,
fn_correct_values['PCA.components_'],
'PCA.components_')
if not does_data_match(fn_return_value.components_,
fn_correct_values['PCA.components_']):
raise Exception('Test Failure: PCA not fitted correctly\n\n'
'PCA.components_:\n'
'{}\n\n'
'Expected PCA.components_:\n'
'{}'.format(fn_return_value.components_,
fn_correct_values['PCA.components_']))
def test_fit_pca(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
'returns': pd.DataFrame(
[
[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734]],
dates, assets),
'num_factor_exposures': 2,
'svd_solver': 'full'}
fn_correct_values = {
'PCA': PCA(),
'PCA.components_': np.array([
[0.81925896, -0.40427891, 0.40666118],
[-0.02011128, 0.68848693, 0.72496985]])}
pca_fit = PCA.fit
with patch.object(PCA, 'fit', autospec=True) as mock_fit:
mock_fit.side_effect = pca_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, fn_correct_values['PCA'], 'PCA')
try:
fn_return_value.fit.assert_called()
except AssertionError:
raise Exception('Test Failure: PCA.fit not called')
try:
fn_return_value.fit.assert_called_with(self=fn_return_value, X=fn_inputs['returns'])
except Exception:
raise Exception('Test Failure: PCA.fit called with the wrong arguments')
assert_structure(fn_return_value.components_, fn_correct_values['PCA.components_'], 'PCA.components_')
if not does_data_match(fn_return_value.components_, fn_correct_values['PCA.components_']):
raise Exception('Test Failure: PCA not fitted correctly\n\n'
'PCA.components_:\n'
'{}\n\n'
'Expected PCA.components_:\n'
'{}'.format(fn_return_value.components_, fn_correct_values['PCA.components_']))
def test_fit_pca(fn):
dates = generate_random_dates(4)
assets = get_assets(3)
fn_inputs = {
"returns":
pd.DataFrame(
[
[0.02769242, 1.34872387, 0.23460972],
[-0.94728692, 0.68386883, -1.23987235],
[1.93769376, -0.48275934, 0.34957348],
[0.23985234, 0.35897345, 0.34598734],
],
dates,
assets,
),
"num_factor_exposures":
2,
"svd_solver":
"full",
}
fn_correct_values = {
"PCA":
PCA(),
"PCA.components_":
np.array([
[0.81925896, -0.40427891, 0.40666118],
[-0.02011128, 0.68848693, 0.72496985],
]),
}
pca_fit = PCA.fit
with patch.object(PCA, "fit", autospec=True) as mock_fit:
mock_fit.side_effect = pca_fit
fn_return_value = fn(**fn_inputs)
assert_structure(fn_return_value, fn_correct_values["PCA"], "PCA")
try:
fn_return_value.fit.assert_called()
except AssertionError:
raise Exception("Test Failure: PCA.fit not called")
try:
fn_return_value.fit.assert_called_with(self=fn_return_value,
X=fn_inputs["returns"])
except Exception:
raise Exception(
"Test Failure: PCA.fit called with the wrong arguments")
assert_structure(
fn_return_value.components_,
fn_correct_values["PCA.components_"],
"PCA.components_",
)
if not does_data_match(fn_return_value.components_,
fn_correct_values["PCA.components_"]):
raise Exception("Test Failure: PCA not fitted correctly\n\n"
"PCA.components_:\n"
"{}\n\n"
"Expected PCA.components_:\n"
"{}".format(fn_return_value.components_,
fn_correct_values["PCA.components_"]))