def test_normalized_variance_derivative(self):
tol = 1.e-8
def compute_der(normvar, sigma):
return (normvar[2:] - normvar[:-2]) / (np.log10(sigma[2:]) -
np.log10(sigma[:-2]))
der, sig = normalized_variance_derivative(self._default_variance_data)
self.assertTrue(
self._default_variance_data.bandwidth_values[1] == sig[0])
self.assertTrue(
self._default_variance_data.bandwidth_values[-2] == sig[-1])
d1 = compute_der(
self._default_variance_data.normalized_variance[self._names[0]],
self._default_variance_data.bandwidth_values)
self.assertTrue(
np.max(np.abs(der[self._names[0]] - d1 / np.max(d1))) <= tol)
# nonzero limit...
indepvars = np.array([[1., 2.], [3., 4.], [1., 2.], [5., 7.]])
depvars = np.array([[2.], [5.], [8.], [10.]])
variance_data = compute_normalized_variance(indepvars, depvars,
self._names)
der, sig = normalized_variance_derivative(variance_data)
d1 = compute_der(variance_data.normalized_variance[self._names[0]],
variance_data.bandwidth_values)
self.assertFalse(
np.max(np.abs(der[self._names[0]] - d1 / np.max(d1))) <= tol)
d1 += variance_data.normalized_variance_limit[self._names[0]]
self.assertTrue(
np.max(np.abs(der[self._names[0]] - d1 / np.max(d1))) <= tol)
def test_bandwidth_array(self):
bw = np.array([0.001, 0.1])
variance_data = compute_normalized_variance(self._indepvars,
self._depvars,
self._names,
bandwidth_values=bw)
self.assertTrue(np.all(variance_data.bandwidth_values == bw))
self.assertTrue(
variance_data.bandwidth_10pct_rise[self._names[0]] is None)
def test_plot_normalized_variance_derivative(self):
X = np.random.rand(100, 5)
pca_X = PCA(X, n_components=2)
principal_components = pca_X.transform(X)
variance_data = compute_normalized_variance(
principal_components,
X,
depvar_names=['A', 'B', 'C', 'D', 'E'],
bandwidth_values=np.logspace(-3, 1, 20),
scale_unit_box=True)
try:
plt = plot_normalized_variance_derivative(variance_data,
plot_variables=[0, 1, 2],
color_map='Blues',
figure_size=(10, 5),
title=None,
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
try:
plt = plot_normalized_variance_derivative(
variance_data,
plot_variables=[],
color_map='Blues',
figure_size=(10, 5),
title='Normalized variance',
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
try:
plt = plot_normalized_variance_derivative(
variance_data,
plot_variables=[2, 3, 4],
color_map='Blues',
figure_size=(15, 5),
title='Normalized variance',
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
try:
plt = plot_normalized_variance_derivative(variance_data,
plot_variables=[],
color_map='Reds',
figure_size=(10, 5),
title=None,
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
def __init__(self, *args, **kwargs):
super(TestNormalizedVariance, self).__init__(*args, **kwargs)
self._indepvars = np.array([[1., 2.], [3., 5.], [6., 9.]])
self._depvars = np.array([[2.], [4.], [1.]])
self._names = ['depvar']
self._default_variance_data = compute_normalized_variance(
self._indepvars, self._depvars, self._names)
self._scl_indepvars = (self._indepvars - np.min(
self._indepvars, axis=0)) / (np.max(self._indepvars, axis=0) -
np.min(self._indepvars, axis=0))
def test_nonzero_normalized_variance_limit(self):
indepvars = np.array([[1., 2.], [3., 4.], [1., 2.]])
depvars = np.array([[2.], [5.], [8.]])
variance_data = compute_normalized_variance(indepvars, depvars,
self._names)
zerotol = 1.e-16
self.assertTrue(
variance_data.normalized_variance_limit[self._names[0]] > zerotol)
self.assertTrue(self._default_variance_data.normalized_variance_limit[
self._names[0]] <= zerotol)
def test_normalized_variance(self):
gs_normvar = 0.69507212
bw = np.array([1.])
variance_data = compute_normalized_variance(self._indepvars,
self._depvars,
self._names,
bandwidth_values=bw)
self.assertTrue(
np.abs(variance_data.normalized_variance[self._names[0]][0] -
gs_normvar) < 1.e-6)
def test_plot_normalized_variance_derivative_comparison(self):
X = np.random.rand(100, 5)
Y = np.random.rand(100, 5)
Z = np.random.rand(100, 5)
pca_X = PCA(X, n_components=2)
pca_Y = PCA(Y, n_components=2)
pca_Z = PCA(Y, n_components=2)
principal_components_X = pca_X.transform(X)
principal_components_Y = pca_Y.transform(Y)
principal_components_Z = pca_Z.transform(Z)
variance_data_X = compute_normalized_variance(
principal_components_X,
X,
depvar_names=['A', 'B', 'C', 'D', 'E'],
bandwidth_values=np.logspace(-3, 2, 20),
scale_unit_box=True)
variance_data_Y = compute_normalized_variance(
principal_components_Y,
Y,
depvar_names=['F', 'G', 'H', 'I', 'J'],
bandwidth_values=np.logspace(-3, 2, 20),
scale_unit_box=True)
variance_data_Z = compute_normalized_variance(
principal_components_Z,
Z,
depvar_names=['K', 'L', 'M', 'N', 'O'],
bandwidth_values=np.logspace(-3, 2, 20),
scale_unit_box=True)
try:
plt = plot_normalized_variance_derivative_comparison(
(variance_data_X, variance_data_Y), ([0, 1, 2], [0, 1, 2]),
('Blues', 'Reds'),
title=None,
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
try:
plt = plot_normalized_variance_derivative_comparison(
(variance_data_X, variance_data_Y), ([0, 1, 2], [0, 1, 2]),
('Blues', 'Reds'),
title='Normalized variance comparison',
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
try:
plt = plot_normalized_variance_derivative_comparison(
(variance_data_X, variance_data_Y, variance_data_Z),
([0, 1, 2], [0, 1, 2], []), ('Greys', 'Blues', 'Reds'),
title='Normalized variance comparison',
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)
try:
plt = plot_normalized_variance_derivative_comparison(
(variance_data_X, variance_data_Y, variance_data_Z),
([0], [2, 3], []), ('Greys', 'Blues', 'Reds'),
title=None,
save_filename=None)
plt.close()
except Exception:
self.assertTrue(False)