def test_OptionValueField_raises_error_when_sum_of_probabilities_is_not_one():
probs = {"A": 0.8, "B": 0.02}
with pytest.raises(ValueError):
field.OptionValueField(probabilities=probs)
probs = {"A": 0.8, "B": 2}
with pytest.raises(ValueError):
field.OptionValueField(probabilities=probs)
def test_OptionValueField_chooses_value_basedon_probability_from_distribution(
monkeypatch):
n = mock.Mock()
mock_choices = {"M": 0.2, "F": 0.8}
n.return_value = "M"
monkeypatch.setattr("random.choice", n)
dob = field.OptionValueField(probabilities=mock_choices)
assert dob.next_value(row) == "M"
n.assert_called_with(dob._option_picks)
def test_OptionValueField_handles_small_probability_when_choosing_options():
probs = {"A": 0.8, "B": 0.02, "C": 0.0001, "D": 0.1799}
ovf = field.OptionValueField(probabilities=probs)
HeadFake.set_seed(1234)
assert len(ovf._option_picks) == 10000
assert len(list(filter(lambda x: x == "A", ovf._option_picks))) == 8000
assert len(list(filter(lambda x: x == "B", ovf._option_picks))) == 200
assert len(list(filter(lambda x: x == "C", ovf._option_picks))) == 1
assert len(list(filter(lambda x: x == "D", ovf._option_picks))) == 1799
assert ["A", "A", "A", "A",
"D"] == [ovf.next_value(row) for i in range(1, 6)]
def test_IfElseField_handles_condition_as_dictionary():
fset = Fieldset(fields={"marital_status": "M"})
HeadFake.set_seed(5)
gender_cond = {"field": "gender", "operator": operator.eq, "value": "M"}
ms_test_cond = {
"field": "marital_status",
"operator": operator.eq,
"value": "M"
}
female_ov = field.OptionValueField(probabilities={
"MISS": 0.7,
"MS": 0.1,
"DR": 0.1,
"PROF": 0.1
})
ms_if_else = field.IfElseField(condition=ms_test_cond,
true_value="MRS",
false_value=female_ov)
gender_if_else = field.IfElseField(condition=gender_cond,
true_value="MR",
false_value=ms_if_else)
gender_if_else.init_from_fieldset(fset)
assert gender_if_else.next_value({
"gender": "M",
"marital_status": "M"
}) == "MR"
assert gender_if_else.next_value({
"gender": "M",
"marital_status": "S"
}) == "MR"
assert gender_if_else.next_value({
"gender": "F",
"marital_status": "S"
}) == "PROF"
assert gender_if_else.next_value({
"gender": "F",
"marital_status": "S"
}) == "MISS"
assert gender_if_else.next_value({
"gender": "F",
"marital_status": "S"
}) == "MISS"
def test_OptionValueField_handles_very_small_probability_with_warning_when_choosing_options(
monkeypatch):
from unittest.mock import Mock
warn = Mock()
monkeypatch.setattr("warnings.warn", warn)
probs = {"A": 0.8, "B": 0.02, "C": 0.000001, "D": 0.179999}
ovf = field.OptionValueField(probabilities=probs)
HeadFake.set_seed(1234)
assert len(ovf._option_picks) == 1000000
assert len(list(filter(lambda x: x == "A", ovf._option_picks))) == 800000
assert len(list(filter(lambda x: x == "B", ovf._option_picks))) == 20000
assert len(list(filter(lambda x: x == "C", ovf._option_picks))) == 1
assert len(list(filter(lambda x: x == "D", ovf._option_picks))) == 179999
assert ["B", "A", "A", "A",
"A"] == [ovf.next_value(row) for i in range(1, 6)]
warn.assert_called_with(
"Options include probabilities of 1e-6. This requires the creation of 1e6 possible options"
)