def test_compute_grouped_elementary_effects():
model_inputs = np.array([[.39, -.39, -1.64, 0.39, -0.39, -0.39, 0.39, 0.39,
-1.64, -0.39, 0.39, -1.64, 1.64, 1.64, 1.64],
[-1.64, 1.64, 0.39, -1.64, 1.64, 1.64, -1.64,
-1.64, -1.64, 1.64, 0.39, -1.64, 1.64, 1.64,
1.64],
[-1.64, 1.64, 0.39, -1.64, 1.64, 1.64, -1.64,
-1.64, 0.39, 1.64, -1.64, 0.39, -.39, -.39, -.39]
])
model_results = np.array([13.85, -10.11, 1.12])
problem = {'names': ['1', '2', '3', '4', '5', '6', '7', '8',
'9', '10', '11', '12', '13', '14', '15'],
'bounds': [[]],
'groups':
(np.array([[0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 1],
[1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 0, 0, 0]]),
['gp1', 'gp2']),
'num_vars': 15
}
ee = compute_elementary_effects(model_inputs, model_results, 3, 2. / 3)
mu_star = np.average(np.abs(ee), axis=1)
actual = compute_grouped_metric(mu_star, problem['groups'][0].T)
desired = np.array([16.86, 35.95])
assert_allclose(actual, desired, atol=1e-1)
def test_compute_grouped_mu_star():
'''
Computes mu_star for 3 variables grouped into 2 groups
There are six trajectories.
'''
group_matrix = np.array([[1, 0], [0, 1], [0, 1]], dtype=int)
ee = np.array([[2.52, 2.01, 2.30, -0.66, -0.93, -1.30],
[-2.00, 0.13, -0.80, 0.25, -0.02, 0.51],
[2.00, -0.13, 0.80, -0.25, 0.02, -0.51]])
mu_star = np.average(np.abs(ee), 1)
actual = compute_grouped_metric(mu_star, group_matrix)
desired = np.array([1.62, 0.62], dtype=float)
assert_allclose(actual, desired, rtol=1e-1)
def test_compute_grouped_mu_star():
'''
Computes mu_star for 3 variables grouped into 2 groups
There are six trajectories.
'''
group_matrix = np.matrix('1,0;0,1;0,1', dtype=np.int)
ee = np.array([[2.52, 2.01, 2.30, -0.66, -0.93, -1.30],
[-2.00, 0.13, -0.80, 0.25, -0.02, 0.51],
[2.00, -0.13, 0.80, -0.25, 0.02, -0.51]])
mu_star = np.average(np.abs(ee), 1)
actual = compute_grouped_metric(mu_star, group_matrix)
desired = np.array([1.62, 0.62], dtype=np.float)
assert_allclose(actual, desired, rtol=1e-1)
