def test_plot_sensitivity(self):
"""Test plot sensitivity indices first oder"""
exp_unc, _, haz_unc = make_imp_uncs()
samples = pd.DataFrame({
'x_exp': [1, 2, 3, 4],
'x_haz': [0.1, 0.2, 0.3, 0.4]
})
metrics = {
'freq_curve':
pd.DataFrame({
'rp100': [9, 10, 11, 12],
'rp250': [100, 110, 120, 130]
})
}
unc = Uncertainty(unc_vars={
'exp': exp_unc,
'haz': haz_unc
},
samples=samples,
metrics=metrics)
unc.calc_sensitivity(method_kwargs={'calc_second_order': False})
unc.plot_sensitivity()
plt.close()
unc.calc_sensitivity(salib_method='rbd_fast', method_kwargs={'M': 8})
unc.plot_sensitivity()
plt.close()
def test_plot_distribution(self):
"""Test plot metrics distribution"""
exp_unc, _, haz_unc = make_imp_uncs()
samples = pd.DataFrame({
'x_exp': [1, 2, 3, 4],
'x_haz': [0.1, 0.2, 0.3, 0.4]
})
metrics = {
'freq_curve':
pd.DataFrame({
'rp100': [9, 10, 11, 12],
'rp250': [100, 110, 120, 130]
})
}
unc = Uncertainty(unc_vars={
'exp': exp_unc,
'haz': haz_unc
},
samples=samples,
metrics=metrics)
unc.plot_distribution()
plt.close()
def test_calc_sensitivty_pass(self):
"""Test compute sensitivity default"""
exp_unc, _, haz_unc = make_imp_uncs()
samples = pd.DataFrame({
'x_exp': [1, 2, 3, 4],
'x_haz': [0.1, 0.2, 0.3, 0.4]
})
metrics = {
'rp':
pd.DataFrame({
'rp100': [9, 10, 11, 12],
'rp250': [100, 110, 120, 130]
})
}
unc = Uncertainty(unc_vars={
'exp': exp_unc,
'haz': haz_unc
},
samples=samples,
metrics=metrics)
sens = unc.calc_sensitivity(method_kwargs={'calc_second_order': False})
self.assertSetEqual(set(sens.keys()), {'rp'})
self.assertSetEqual(set(sens['rp'].keys()), {'rp100', 'rp250'})
self.assertSetEqual(set(sens['rp']['rp100'].keys()),
{'S1', 'S1_conf', 'ST', 'ST_conf'})
self.assertTrue(
np.allclose(sens['rp']['rp100']['S1'],
np.array([0.66666667, 1.33333333])))
def test_calc_sensitivty_XY_pass(self):
"""Test compute sensitvity method rbd_fast (variables names different
from default)"""
exp_unc, _, haz_unc = make_imp_uncs()
samples = pd.DataFrame({
'x_exp': [1, 2, 3, 4],
'x_haz': [0.1, 0.2, 0.3, 0.4]
})
metrics = {
'rp':
pd.DataFrame({
'rp100': [9.0, 10.0, 11.0, 12.0],
'rp250': [100.0, 110.0, 120.0, 130.0]
})
}
unc = Uncertainty(unc_vars={
'exp': exp_unc,
'haz': haz_unc
},
samples=samples,
metrics=metrics)
sens = unc.calc_sensitivity(salib_method='rbd_fast',
method_kwargs={'M': 8})
self.assertSetEqual(set(sens.keys()), {'rp'})
self.assertSetEqual(set(sens['rp'].keys()), {'rp100', 'rp250'})
self.assertSetEqual(set(sens['rp']['rp100'].keys()), {'S1', 'names'})
self.assertTrue(
np.allclose(sens['rp']['rp100']['S1'], np.array([1.0, 1.0])))
def test_init_pass(self):
"""Test initiliazation uncertainty"""
exp_unc, impf_unc, haz_unc = make_imp_uncs()
unc = Uncertainty({'exp': exp_unc, 'impf': impf_unc, 'haz': haz_unc})
self.assertDictEqual(unc.metrics, {})
self.assertDictEqual(unc.sensitivity, {})
self.assertEqual(unc.n_samples, 0)
self.assertSetEqual(set(unc.param_labels),
{'x_exp', 'x_haz', 'x_paa', 'x_mdd'})
self.assertSetEqual(set(unc.problem_sa['names']),
{'x_exp', 'x_haz', 'x_paa', 'x_mdd'})
self.assertSetEqual(set(unc.distr_dict.keys()),
{"x_exp", "x_paa", "x_mdd", "x_haz"})
unc = Uncertainty(
{
'exp': exp_unc,
'impf': impf_unc
},
samples=pd.DataFrame({
'x_exp': [1, 2],
'x_paa': [3, 4],
'x_mdd': [1, 2]
}),
metrics={'aai_agg': pd.DataFrame({'aai_agg': [100, 200]})})
self.assertEqual(unc.n_samples, 2)
self.assertSetEqual(set(unc.param_labels), {'x_exp', 'x_paa', 'x_mdd'})
self.assertListEqual(list(unc.metrics['aai_agg']['aai_agg']),
[100, 200])
self.assertDictEqual(unc.sensitivity, {})
def test_save_pass(self):
"""Test save samples"""
exp_unc, impf_unc, haz_unc = make_imp_uncs()
unc = Uncertainty({'exp': exp_unc, 'impf': impf_unc, 'haz': haz_unc})
unc.make_sample(1)
filename = unc.save_samples_df()
unc_imp = UncImpact(exp_unc, impf_unc, haz_unc)
unc_imp.load_samples_df(filename)
unc_imp.calc_distribution()
unc_imp.calc_sensitivity()
def test_make_sample_pass(self):
"""Test generate sample"""
exp_unc, _, haz_unc = make_imp_uncs()
unc = Uncertainty({'exp': exp_unc, 'haz': haz_unc})
#default sampling saltelli
unc.make_sample(N=1, sampling_kwargs={'calc_second_order': True})
self.assertEqual(unc.n_samples, 1 * (2 * 2 + 2)) # N * (2 * D + 2)
self.assertTrue(isinstance(unc.samples_df, pd.DataFrame))
self.assertTrue(
np.allclose(unc.samples_df['x_exp'],
np.array([
1.239453, 1.837109, 1.239453, 1.239453, 1.837109,
1.837109
]),
rtol=1e-05))
self.assertListEqual(list(unc.samples_df['x_haz']),
[0.0, 0.0, 1.0, 1.0, 0.0, 1.0])
#latin sampling
unc.make_sample(N=1,
sampling_method='latin',
sampling_kwargs={'seed': 11245})
self.assertEqual(unc.n_samples, 1)
self.assertTrue(isinstance(unc.samples_df, pd.DataFrame))
self.assertTrue(
np.allclose(unc.samples_df['x_exp'],
np.array([2.58309]),
rtol=1e-05))
self.assertListEqual(list(unc.samples_df['x_haz']), [2.0])
def test_plot_sample_pass(self):
exp_unc, _, haz_unc = make_imp_uncs()
unc = Uncertainty({'exp': exp_unc, 'haz': haz_unc})
unc.make_sample(N=1)
unc.plot_sample()
plt.close()
def test_est_comp_time_pass(self):
exp_unc, _, haz_unc = make_imp_uncs()
unc = Uncertainty({'exp': exp_unc, 'haz': haz_unc})
unc.make_sample(N=1, sampling_kwargs={'calc_second_order': False})
est = unc.est_comp_time(0.12345)
self.assertEqual(est, 1 * (2 + 2) * 0.123) # N * (D + 2)
pool = Pool(nodes=4)
est = unc.est_comp_time(0.12345, pool)
self.assertEqual(est, 1 * (2 + 2) * 0.123 / 4) # N * (D + 2)
pool.close()
pool.join()
pool.clear()