def test__has_not_count_errors(self):
# Test unsupported dtype series is given
start_time = self.example_data_A_data['start_time']
start_time = start_time.astype(np.datetime64)
# Test has version
with self.assertRaises(TypeError) as error:
test_values = [1, 2]
# Test _has_count raises TypeError
idx = _has_count(start_time, test_values)
self.assertEqual(
error.exception.message[:58],
"The series given to has_count() must be a supported dtype."
)
# Test not version
with self.assertRaises(TypeError) as error:
test_values = [1, 2]
# Test _has_count raises TypeError
idx = _not_count(start_time, test_values)
self.assertEqual(
error.exception.message[:58],
"The series given to not_count() must be a supported dtype."
)
def test__has_not_count_errors(self):
# Test unsupported dtype series is given
start_time = self.example_data_A_data['start_time']
start_time = start_time.astype(np.datetime64)
# Test has version
with self.assertRaises(TypeError) as error:
test_values = [1, 2]
# Test _has_count raises TypeError
idx = _has_count(start_time, test_values)
self.assertEqual(
error.exception.message[:58],
"The series given to has_count() must be a supported dtype."
)
# Test not version
with self.assertRaises(TypeError) as error:
test_values = [1, 2]
# Test _has_count raises TypeError
idx = _not_count(start_time, test_values)
self.assertEqual(
error.exception.message[:58],
"The series given to not_count() must be a supported dtype."
)
def test__has_not_count(self):
test_vars = [
'q2', # Test on delimited set
'gender', # Test on single stored as int64
'locality' # Test on single stored as float64
]
# Test non-operator-lead logical comparisons
for var_name in test_vars:
test_var = self.example_data_A_data[var_name]
response_tests = [
[1],
[1, 2],
[1, 3],
[1, 3, [1, 2, 3]]
]
for test_responses in response_tests:
# Test _has_count returns correct results
idx = _has_count(test_var, test_responses)
# Determine min/max values for logic and
# slice dummies column-wise for targeted values subset
dummies, _min, _max = self.get_count_nums(
test_var[idx],
test_responses
)
# Count the number of resposnes per row
test_var_counts = dummies.sum(axis=1).unique()
if len(test_responses)==1:
# Test single targeted response count
self.assertEqual(test_var_counts, [_min])
else:
value_range = range(_min, _max+1)
# Positive test range of response count
self.assertTrue(all([
c in value_range
for c in test_var_counts
]))
# Test inverse index produced by not version
not_idx = _not_count(test_var, test_responses)
self.confirm_inverse_index(test_var, idx, not_idx)
# Test operator-lead logical comparisons
__op_symbol__ = {
_is_lt: '<', _is_le: '<=',
_is_eq: '', _is_ne: '!=',
_is_ge: '>=', _is_gt: '>'
}
__op_map__ = {
_is_lt: lt, _is_le: le,
_is_eq: eq, _is_ne: ne,
_is_ge: ge, _is_gt: gt
}
for op_func in [_is_lt, _is_le, _is_eq, _is_ne, _is_ge, _is_gt]:
key_part = __op_symbol__[op_func]
for var_name in test_vars:
test_var = self.example_data_A_data[var_name]
response_tests = [
[(op_func, 3)],
[(op_func, 3), [1, 2, 3]]
]
for test_responses in response_tests:
# Test _has_count returns correct results
idx = _has_count(test_var, test_responses)
# Determine min/max values for logic and
# slice dummies column-wise for targeted values subset
dummies, dum_func, _max = self.get_count_nums(
test_var[idx],
test_responses
)
numerator = dum_func[1]
try:
values = test_responses[1]
values = [str(v) for v in values if str(v) in dummies.columns]
dummies = dummies[values]
except:
pass
# Count the number of resposnes per row
test_var_counts = dummies.sum(axis=1).unique()
# Positive test range of response count
self.assertTrue(all(__op_map__[op_func](
test_var_counts,
numerator
)))
if op_func in [_is_ge, _is_eq] and numerator > 0:
incl_na = False
elif op_func in [_is_gt]:
incl_na = False
else:
incl_na = True
# Test inverse index produced by not version
not_idx = _not_count(test_var, test_responses)
self.confirm_inverse_index(
test_var,
idx,
not_idx,
incl_na
)
# Test non-operator-lead logical comparisons with
# exclusivity
for var_name in test_vars:
test_var = self.example_data_A_data[var_name]
response_tests = [
[1, 3, [1, 2, 3]]
]
for test_responses in response_tests:
# Test _has_count returns correct results
idx = _has_count(test_var, test_responses, True)
# Determine min/max values for logic and
# slice dummies column-wise for targeted values subset
dummies, _min, _max = self.get_count_nums(
test_var[idx],
test_responses
)
# Count the number of resposnes per row
test_var_counts = dummies.sum(axis=1).unique()
value_range = range(_min, _max+1)
# Positive test range of response count
self.assertTrue(all([
c in value_range
for c in test_var_counts
]))
# Negative test for exclusivity
all_dummies = test_var.astype('object').str.get_dummies(';')
other_cols = [
c for c in all_dummies.columns
if not c in dummies.columns
]
other_dummies = all_dummies[other_cols]
other_any_mask = other_dummies.any(axis=1)
other_dummies = other_dummies[other_any_mask]
self.assertEqual(
other_dummies.index.intersection(dummies.index).size,
0
)
# Test inverse index produced by not version
not_idx = _not_count(test_var, test_responses, True)
self.confirm_inverse_index(test_var, idx, not_idx)
def test__has_not_count(self):
test_vars = [
'q2', # Test on delimited set
'gender', # Test on single stored as int64
'locality' # Test on single stored as float64
]
# Test non-operator-lead logical comparisons
for var_name in test_vars:
test_var = self.example_data_A_data[var_name]
response_tests = [
[1],
[1, 2],
[1, 3],
[1, 3, [1, 2, 3]]
]
for test_responses in response_tests:
# Test _has_count returns correct results
idx = _has_count(test_var, test_responses)
# Determine min/max values for logic and
# slice dummies column-wise for targeted values subset
dummies, _min, _max = self.get_count_nums(
test_var[idx],
test_responses
)
# Count the number of resposnes per row
test_var_counts = dummies.sum(axis=1).unique()
if len(test_responses)==1:
# Test single targeted response count
self.assertEqual(test_var_counts, [_min])
else:
value_range = range(_min, _max+1)
# Positive test range of response count
self.assertTrue(all([
c in value_range
for c in test_var_counts
]))
# Test inverse index produced by not version
not_idx = _not_count(test_var, test_responses)
self.confirm_inverse_index(test_var, idx, not_idx)
# Test operator-lead logical comparisons
__op_symbol__ = {
_is_lt: '<', _is_le: '<=',
_is_eq: '', _is_ne: '!=',
_is_ge: '>=', _is_gt: '>'
}
__op_map__ = {
_is_lt: lt, _is_le: le,
_is_eq: eq, _is_ne: ne,
_is_ge: ge, _is_gt: gt
}
for op_func in [_is_lt, _is_le, _is_eq, _is_ne, _is_ge, _is_gt]:
key_part = __op_symbol__[op_func]
for var_name in test_vars:
test_var = self.example_data_A_data[var_name]
response_tests = [
[(op_func, 3)],
[(op_func, 3), [1, 2, 3]]
]
for test_responses in response_tests:
# Test _has_count returns correct results
idx = _has_count(test_var, test_responses)
# Determine min/max values for logic and
# slice dummies column-wise for targeted values subset
dummies, dum_func, _max = self.get_count_nums(
test_var[idx],
test_responses
)
numerator = dum_func[1]
try:
values = test_responses[1]
values = [str(v) for v in values if str(v) in dummies.columns]
dummies = dummies[values]
except:
pass
# Count the number of resposnes per row
test_var_counts = dummies.sum(axis=1).unique()
# Positive test range of response count
self.assertTrue(all(__op_map__[op_func](
test_var_counts,
numerator
)))
if op_func in [_is_ge, _is_eq] and numerator > 0:
incl_na = False
elif op_func in [_is_gt]:
incl_na = False
else:
incl_na = True
# Test inverse index produced by not version
not_idx = _not_count(test_var, test_responses)
self.confirm_inverse_index(
test_var,
idx,
not_idx,
incl_na
)
# Test non-operator-lead logical comparisons with
# exclusivity
for var_name in test_vars:
test_var = self.example_data_A_data[var_name]
response_tests = [
[1, 3, [1, 2, 3]]
]
for test_responses in response_tests:
# Test _has_count returns correct results
idx = _has_count(test_var, test_responses, True)
# Determine min/max values for logic and
# slice dummies column-wise for targeted values subset
dummies, _min, _max = self.get_count_nums(
test_var[idx],
test_responses
)
# Count the number of resposnes per row
test_var_counts = dummies.sum(axis=1).unique()
value_range = range(_min, _max+1)
# Positive test range of response count
self.assertTrue(all([
c in value_range
for c in test_var_counts
]))
# Negative test for exclusivity
all_dummies = test_var.astype('object').str.get_dummies(';')
other_cols = [
c for c in all_dummies.columns
if not c in dummies.columns
]
other_dummies = all_dummies[other_cols]
other_any_mask = other_dummies.any(axis=1)
other_dummies = other_dummies[other_any_mask]
self.assertEqual(
other_dummies.index.intersection(dummies.index).size,
0
)
# Test inverse index produced by not version
not_idx = _not_count(test_var, test_responses, True)
self.confirm_inverse_index(test_var, idx, not_idx)