def test_add_noise(self):
noise = 1
params = OrderedDict([('loc', 0), ('scale', 1)])
np.random.seed(31337)
new_params = _add_noise({'test': params}, noise=noise)['test']
# check if copied
self.assertFalse(new_params is params)
# check if noise correctly added
np.random.seed(31337)
self.assertTrue(new_params['loc'] == np.random.normal(loc=params['loc'], scale=noise))
self.assertTrue(new_params['scale'] == np.random.normal(loc=params['scale'], scale=noise))
# test whether exclude works
new_params = _add_noise({'test': params}, noise=noise, exclude_params=('scale',))['test']
self.assertTrue(new_params['scale'] == 1)
# test whether bounds work
for i in range(10):
bound_params = _add_noise({'test': params}, bounds={'loc': (-1, 1), 'scale': (0, 2)}, noise=3)['test']
assert (bound_params['loc'] > -1) and (bound_params['loc'] < 1)
assert (bound_params['scale'] > 0) and (bound_params['scale'] < 2)
# test whether valid_func works
check_valid_func = lambda **params: (params['loc'] > -1) and (params['loc'] < 1) and (params['scale'] > 0) and (params['scale'] < 2)
for i in range(10):
bound_params = _add_noise({'test': params}, check_valid_func=check_valid_func, noise=3)['test']
assert (bound_params['loc'] > -1) and (bound_params['loc'] < 1)
assert (bound_params['scale'] > 0) and (bound_params['scale'] < 2)
def test_single_cond_multi_subjs(self):
params = OrderedDict([('loc', 0), ('scale', 1)])
subjs = 3
size = 100
# generate test data
seed = 31337
data, params_subjs = gen_rand_data(gen_func_df,
params,
size=size,
subjs=subjs,
seed=seed)
# test subj present
np.testing.assert_array_equal(np.unique(data['subj_idx']),
list(range(subjs)))
# test for correct length
np.testing.assert_array_equal(len(data), subjs * size)
# generate truth
np.random.seed(seed)
for i in range(subjs):
new_params = _add_noise({'test': params})['test']
print("check", new_params)
truth = gen_func_df(size=size, **new_params)
np.testing.assert_array_equal(data[data['subj_idx'] == i]['data'],
truth['data'])
self.assertEqual(params_subjs[i], new_params)
def test_add_noise(self):
noise = 1
params = OrderedDict([("loc", 0), ("scale", 1)])
np.random.seed(31337)
new_params = _add_noise(params, noise=noise)
# check if copied
self.assertFalse(new_params is params)
# check if noise correctly added
np.random.seed(31337)
self.assertTrue(new_params["loc"] == np.random.normal(loc=params["loc"], scale=noise))
self.assertTrue(new_params["scale"] == np.random.normal(loc=params["scale"], scale=noise))
# test whether exclude works
new_params = _add_noise(params, noise=noise, exclude_params=("scale",))
self.assertTrue(new_params["scale"] == 1)
def test_add_noise(self):
noise = 1
params = OrderedDict([('loc', 0), ('scale', 1)])
np.random.seed(31337)
new_params = _add_noise({'test': params}, noise=noise)['test']
# check if copied
self.assertFalse(new_params is params)
# check if noise correctly added
np.random.seed(31337)
self.assertTrue(new_params['loc'] == np.random.normal(
loc=params['loc'], scale=noise))
self.assertTrue(new_params['scale'] == np.random.normal(
loc=params['scale'], scale=noise))
# test whether exclude works
new_params = _add_noise({'test': params},
noise=noise,
exclude_params=('scale', ))['test']
self.assertTrue(new_params['scale'] == 1)
# test whether bounds work
for i in range(10):
bound_params = _add_noise({'test': params},
bounds={
'loc': (-1, 1),
'scale': (0, 2)
},
noise=3)['test']
assert (bound_params['loc'] > -1) and (bound_params['loc'] < 1)
assert (bound_params['scale'] > 0) and (bound_params['scale'] < 2)
# test whether valid_func works
check_valid_func = lambda **params: (params['loc'] > -1) and (params[
'loc'] < 1) and (params['scale'] > 0) and (params['scale'] < 2)
for i in range(10):
bound_params = _add_noise({'test': params},
check_valid_func=check_valid_func,
noise=3)['test']
assert (bound_params['loc'] > -1) and (bound_params['loc'] < 1)
assert (bound_params['scale'] > 0) and (bound_params['scale'] < 2)
def test_single_cond_multi_subjs_exclude(self):
params = OrderedDict([("loc", 0), ("scale", 1)])
subjs = 100
samples = 100
# generate test data
np.random.seed(31337)
data, params_subjs = gen_rand_data(normal_like, params, samples=samples, subjs=subjs, exclude_params=("scale",))
# test subj present
np.testing.assert_array_equal(np.unique(data["subj_idx"]), range(subjs))
# test for correct length
np.testing.assert_array_equal(len(data), subjs * samples)
# generate truth
np.random.seed(31337)
for i in range(subjs):
new_params = _add_noise(params, exclude_params=("scale",))
truth = np.float64(normal_like.rv.random(size=samples, **new_params))
np.testing.assert_array_equal(data[data["subj_idx"] == i]["data"], truth)
self.assertEqual(params_subjs[i], new_params)
def test_single_cond_multi_subjs_exclude(self):
params = OrderedDict([('loc', 0), ('scale', 1)])
subjs = 3
size = 100
# generate test data
seed = 31337
data, params_subjs = gen_rand_data(gen_func_df, params, size=size, subjs=subjs,
exclude_params=('scale',), seed=seed)
# test subj present
np.testing.assert_array_equal(np.unique(data['subj_idx']), range(subjs))
# test for correct length
np.testing.assert_array_equal(len(data), subjs*size)
# generate truth
np.random.seed(seed)
for i in range(subjs):
new_params = _add_noise({'test': params}, exclude_params=('scale',))['test']
truth = gen_func_df(size=size, **new_params)
np.testing.assert_array_equal(data[data['subj_idx'] == i]['data'], truth['data'])
self.assertEqual(params_subjs[i], new_params)
