def test_PopClass_get_conditional_population():
"""
The get_conditional_population() method should return the correct
population when supplied with a certain conditions file.
"""
# Define test population
popsize = 100
random_seed = 100
pop_class = PopulationClass(popsize, random_seed)
yaml_folder = "./tests/testdata/PopulationClass/"
yaml_filenames = find_yamls(yaml_folder)
yaml_objects = load_yamls(yaml_filenames)
probab_objects = []
for y_obj in yaml_objects:
if y_obj["data_type"] in ["categorical", "ordinal"]:
probab_objects.append(DiscreteProbabilityClass(y_obj))
elif y_obj["data_type"] in ["continuous"]:
probab_objects.append(ContinuousProbabilityClass(y_obj))
for probab_object in probab_objects:
pop_class.add_property(probab_object)
pop_class.update(property_name="all")
population_cond = pop_class.get_conditional_population('age', cond_index=0)
test_population_cond = pop_class.population.query("sex == 0")
test_population_cond = test_population_cond[['person_id']]
assert population_cond.equals(test_population_cond)
population_cond = pop_class.get_conditional_population('age', cond_index=1)
test_population_cond = pop_class.population.query("sex == 1")
test_population_cond = test_population_cond[['person_id']]
assert population_cond.equals(test_population_cond)
def test_generate_population():
"""
Test the function generate_population()
Test for an integer random seed and for a random seed equal to None.
"""
# Define test population
popsize = 100
random_seed = 100
test_pop_class = PopulationClass(popsize, random_seed)
yaml_folder = "./tests/testdata/PopulationClass/"
yaml_filenames = find_yamls(yaml_folder)
yaml_objects = load_yamls(yaml_filenames)
probab_objects = []
for y_obj in yaml_objects:
if y_obj["data_type"] in ["categorical", "ordinal"]:
probab_objects.append(DiscreteProbabilityClass(y_obj))
elif y_obj["data_type"] in ["continuous"]:
probab_objects.append(ContinuousProbabilityClass(y_obj))
for probab_object in probab_objects:
test_pop_class.add_property(probab_object)
pop_class = generate_population(popsize,
yaml_folder,
rand_seed=random_seed)
assert test_pop_class == pop_class
# Test for random_seed=None
test_pop_class_no_seed = PopulationClass(popsize)
for probab_object in probab_objects:
test_pop_class_no_seed.add_property(probab_object)
pop_class_no_seed = generate_population(popsize, yaml_folder)
assert pop_class_no_seed == test_pop_class_no_seed
def test_load_yamls_multi_prop():
"""
If multiple YAML files are defined for the same property, trigger
an AssertionError.
"""
testfolder = "./tests/testdata/load_yamls-multiple_properties/"
yaml_filenames = find_yamls(testfolder)
with pytest.raises(AssertionError):
yaml_objects = load_yamls(yaml_filenames)
del yaml_objects
def test_find_yamls():
"""
Check if find_yamls() only finds the files with the correct filename
format.
"""
folder_name = "./tests/testdata/find_yamls/"
yaml_filenames = find_yamls(folder_name)
test_filenames = [
"./tests/testdata/find_yamls/a.yml",
"./tests/testdata/find_yamls/b.yaml",
]
assert yaml_filenames == test_filenames
def test_PopClass_export():
"""
Test the export method.
- Test the assertions that check the argument types.
"""
test_output_filename = "./tests/testdata/PopulationClass/export.csv"
if os.path.exists(test_output_filename):
os.remove(test_output_filename)
# Define test population
popsize = 100
random_seed = 100
pop_class = PopulationClass(popsize, random_seed)
yaml_folder = "./tests/testdata/PopulationClass/"
yaml_filenames = find_yamls(yaml_folder)
yaml_objects = load_yamls(yaml_filenames)
probab_objects = []
for y_obj in yaml_objects:
if y_obj["data_type"] in ["categorical", "ordinal"]:
probab_objects.append(DiscreteProbabilityClass(y_obj))
elif y_obj["data_type"] in ["continuous"]:
probab_objects.append(ContinuousProbabilityClass(y_obj))
for probab_object in probab_objects:
pop_class.add_property(probab_object)
pop_class.update(property_name="all")
# Test the assertions that check the argument types.
with pytest.raises(AssertionError):
pop_class.export(output=0)
with pytest.raises(AssertionError):
pop_class.export(output=test_output_filename, nowrite="notaboolean")
pop_class.export(test_output_filename, nowrite=True)
assert not os.path.exists(test_output_filename), \
"Export with nowrite writes to file"
pop_class.export(test_output_filename)
pop_class_from_export = pd.read_csv(test_output_filename, sep=',')
population_w_labels = pop_class.population.copy()
for prob_obj in pop_class.prob_objects.values():
if prob_obj.data_type in ["categorical", "ordinal"]:
prop = prob_obj.property_name
population_w_labels[prop] = population_w_labels[prop]\
.apply(lambda idx: 'nodata' if np.isnan(idx)
else prob_obj.labels[int(idx)])
assert_frame_equal(pop_class_from_export, population_w_labels), \
"Population written to file is different than calculated"
def test_PopClass_update():
"""
When updating the PopulationClass a new column of each
property should be added on the first time updating that
property. Any subsequent times the columns should remain the same.
The age column should be completely filled.
"""
popsize = 10
random_seed = 100
pop_class = PopulationClass(popsize, random_seed)
yaml_folder = "./tests/testdata/PopulationClass/"
yaml_filenames = find_yamls(yaml_folder)
yaml_objects = load_yamls(yaml_filenames)
probab_objects = []
for y_obj in yaml_objects:
if y_obj["data_type"] in ["categorical", "ordinal"]:
probab_objects.append(DiscreteProbabilityClass(y_obj))
elif y_obj["data_type"] in ["continuous"]:
probab_objects.append(ContinuousProbabilityClass(y_obj))
for probab_object in probab_objects:
pop_class.add_property(probab_object)
# No properties yet filled with drawn values
assert sorted(pop_class.population.columns.values) == \
['person_id']
pop_class.update(property_name="sex")
assert sorted(pop_class.population.columns.values) == \
['person_id', 'sex']
pop_class.update(property_name="sex")
assert sorted(pop_class.population.columns.values) == \
['person_id', 'sex']
pop_class.update(property_name="age")
assert sorted(pop_class.population.columns.values) == \
['age', 'person_id', 'sex']
pop_class.update(property_name="income")
assert sorted(pop_class.population.columns.values) == \
['age', 'income', 'person_id', 'sex']
# If all the variables are updated again, no additional columns should
# be created.
pop_class.update(property_name="all")
assert sorted(pop_class.population.columns.values) == \
['age', 'income', 'person_id', 'sex']
def test_PopClass_add_property():
"""
Adding a property with PopulationClass.add_property must work correcly if
the input is correct, a ProbabilityClass is entered as a parameter and
it does not yet exist.
An AssertionError must be raised if
- a non-ProbabilityClass object is entered.
- the property is already defined.
"""
# Define test population
popsize = 100
random_seed = 100
pop_class = PopulationClass(popsize, random_seed)
# Check if pop_class.prob_objects is an empty dictionary at the start
assert isinstance(pop_class.prob_objects, dict) and not \
bool(pop_class.prob_objects)
# Define test ProbabilityClass objects
test_prob_objects = {}
yaml_folder = "./tests/testdata/PopulationClass/"
yaml_filenames = find_yamls(yaml_folder)
yaml_objects = load_yamls(yaml_filenames)
probab_objects = []
for y_obj in yaml_objects:
if y_obj["data_type"] in ["categorical", "ordinal"]:
probab_objects.append(DiscreteProbabilityClass(y_obj))
elif y_obj["data_type"] in ["continuous"]:
probab_objects.append(ContinuousProbabilityClass(y_obj))
for probab_object in probab_objects:
pop_class.add_property(probab_object)
test_prob_objects[probab_object.property_name] = \
probab_object
# Test behaviour for correct input
assert pop_class.prob_objects == test_prob_objects
# Test if prob_object is not a ProbabilityClass object
pop_class.add_property("new_variable")
assert pop_class.prob_objects == test_prob_objects
# Test if prob_object already exists
pop_class.add_property(probab_objects[0])
assert pop_class.prob_objects == test_prob_objects
def test_PopClass_remove_property():
"""
Removing a property with PopulationClass.remove_property must work
correcly if the name of an existing property is entered.
An AssertionError must be raised if
- the entered argument is not a string
- the property is not defined.
"""
# Define test population
popsize = 100
random_seed = 100
pop_class = PopulationClass(popsize, random_seed)
test_prob_objects = {}
yaml_folder = "./tests/testdata/PopulationClass/"
yaml_filenames = find_yamls(yaml_folder)
yaml_objects = load_yamls(yaml_filenames)
probab_objects = []
for y_obj in yaml_objects:
if y_obj["data_type"] in ["categorical", "ordinal"]:
probab_objects.append(DiscreteProbabilityClass(y_obj))
elif y_obj["data_type"] in ["continuous"]:
probab_objects.append(ContinuousProbabilityClass(y_obj))
for probab_object in probab_objects:
pop_class.add_property(probab_object)
test_prob_objects[probab_object.property_name] = \
probab_object
del test_prob_objects['age']
# Test behaviour for correct input
pop_class.remove_property('age')
assert pop_class.prob_objects == test_prob_objects
# Test if property is not in the ProbabilityClass object
pop_class.remove_property('age')
assert pop_class.prob_objects == test_prob_objects
# Test if a non-string is entered
pop_class.remove_property(0)
assert pop_class.prob_objects == test_prob_objects