def test_shuffle_with_random_state():
gm_1 = GaussianMixture(
20,
centers,
covariances,
class_probs,
random_state=42,
shuffle=True,
shuffle_random_state=42,
)
gm_1.sample_views("poly")
Xs_1, y_1 = gm_1.get_Xy()
gm_2 = GaussianMixture(
20,
centers,
covariances,
class_probs,
random_state=42,
shuffle=True,
shuffle_random_state=42,
)
gm_2.sample_views("poly")
Xs_2, y_2 = gm_2.get_Xy()
for view1, view2 in zip(Xs_1, Xs_2):
assert np.allclose(view1, view2)
assert np.allclose(y_1, y_2)
def test_random_state():
gm_1 = GaussianMixture(mu, sigma, 10, class_probs, random_state=42)
gm_1.sample_views('poly')
Xs_1, y_1 = gm_1.get_Xy()
gm_2 = GaussianMixture(mu, sigma, 10, class_probs, random_state=42)
gm_2.sample_views('poly')
Xs_2, y_2 = gm_2.get_Xy()
for view1, view2 in zip(Xs_1, Xs_2):
assert np.allclose(view1, view2)
assert np.allclose(y_1, y_2)
def test_noise_dims_not_same_but_reproducible():
gm_1 = GaussianMixture(mu, sigma, 20, class_probs, random_state=42)
gm_1.sample_views('poly', n_noise=2)
Xs_2, y_2 = gm_1.get_Xy()
view1_noise, view2_noise = Xs_2[0][:, -2:], Xs_2[1][:, -2:]
assert not np.allclose(view1_noise, view2_noise)
gm_2 = GaussianMixture(mu, sigma, 20, class_probs, random_state=42)
gm_2.sample_views('poly', n_noise=2)
Xs_2, y_2 = gm_1.get_Xy()
view1_noise2, view2_noise2 = Xs_2[0][:, -2:], Xs_2[1][:, -2:]
assert np.allclose(view1_noise, view1_noise2)
assert np.allclose(view2_noise, view2_noise2)
def test_noise_dims_not_same_but_reproducible():
gm_1 = GaussianMixture(20,
centers,
covariances,
class_probs,
random_state=42)
gm_1.sample_views("poly", n_noise=2)
Xs_2, _ = gm_1.get_Xy()
view1_noise, view2_noise = Xs_2[0][:, -2:], Xs_2[1][:, -2:]
assert not np.allclose(view1_noise, view2_noise)
gm_2 = GaussianMixture(20,
centers,
covariances,
class_probs,
random_state=42)
gm_2.sample_views("poly", n_noise=2)
Xs_2, _ = gm_1.get_Xy()
view1_noise2, view2_noise2 = Xs_2[0][:, -2:], Xs_2[1][:, -2:]
assert np.allclose(view1_noise, view1_noise2)
assert np.allclose(view2_noise, view2_noise2)
def test_shuffle():
np.random.seed(42)
gm_1 = GaussianMixture(mu,
sigma,
20,
class_probs,
random_state=42,
shuffle=True,
shuffle_random_state=42)
gm_1.sample_views('poly')
Xs_1, y_1 = gm_1.get_Xy()
np.random.seed(30)
gm_2 = GaussianMixture(mu,
sigma,
20,
class_probs,
random_state=42,
shuffle=True,
shuffle_random_state=10)
gm_2.sample_views('poly')
Xs_2, y_2 = gm_2.get_Xy()
for view1, view2 in zip(Xs_1, Xs_2):
assert not np.allclose(view1, view2)
assert not np.allclose(y_1, y_2)
import numpy as np
# Latent variables are sampled from two multivariate Gaussians with equal
# prior probability. Then a polynomial transformation is applied and noise is
# added independently to both the transformed and untransformed latents.
n_samples = 100
centers = [[0, 1], [0, -1]]
covariances = [np.eye(2), np.eye(2)]
gm = GaussianMixture(n_samples,
centers,
covariances,
random_state=42,
shuffle=True,
shuffle_random_state=42)
gm = gm.sample_views(transform='poly', n_noise=2)
latent, y = gm.get_Xy(latents=True)
Xs, _ = gm.get_Xy(latents=False)
# The latent data is plotted against itself to reveal the underlying
# distribtution.
crossviews_plot([latent, latent],
labels=y,
title='Latent Variable',
equal_axes=True)
# The noisy latent variable (view 1) is plotted against the transformed latent
# variable (view 2), an example of a dataset with two views.
n_samples = 200
centers = [[0, 1], [0, -1]]
covariances = 2 * np.array([np.eye(2), np.eye(2)])
gm_train = GaussianMixture(n_samples, centers, covariances)
# Test
gm_test = GaussianMixture(n_samples, centers, covariances)
# Make 2 views
n_noise = 2
transforms = ['linear', 'poly', 'sin']
Xs_train = []
Xs_test = []
for transform in transforms:
gm_train.sample_views(transform=transform, n_noise=n_noise)
gm_test.sample_views(transform=transform, n_noise=n_noise)
Xs_train.append(gm_train.get_Xy()[0])
Xs_test.append(gm_test.get_Xy()[0])
# Plotting parameters
labels = gm_test.latent_[:, 0]
cmap = matplotlib.colors.ListedColormap(
sns.diverging_palette(240, 10, n=len(labels), center='light').as_hex())
cmap = 'coolwarm'
method_labels = \
['Raw Views', 'Linear KCCA', 'Polynomial KCCA', 'Gaussian KCCA', 'DCCA']
transform_labels = \
['Linear Transform', 'Polynomial Transform', 'Sinusoidal Transform']
gm = GaussianMixture(n_samples,
means,
covariances,
random_state=42,
shuffle=True,
shuffle_random_state=42)
latent, y = gm.get_Xy(latents=True)
# Plot latent data against itself to reveal the underlying distribtution.
crossviews_plot([latent, latent],
labels=y,
title='Latent Variable',
equal_axes=True)
# Split data into train and test sets
Xs, y = gm.sample_views(transform='poly', n_noise=2).get_Xy()
Xs_train, Xs_test, y_train, y_test = train_test_split(Xs,
y,
test_size=0.3,
random_state=42)
# Plot the testing data after polynomial transformation
crossviews_plot(Xs_test,
labels=y_test,
title='Testing Data View 1 vs. View 2 '
'(Polynomial Transform + noise)',
equal_axes=True)
###############################################################################
# Fit DCCA model to uncover latent distribution
# ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
======================================
Plotting multiview data with crossplot
======================================
In many cases with multi-view data, especially after use of an embedding
algorithm, one is interested in visualizing two views across dimensions.
One use is assessing correlation between corresponding dimensions of views.
Here, we use this function to display the relationship between two views
simulated from transformations of multi-variant gaussians.
"""
from mvlearn.datasets import GaussianMixture
from mvlearn.plotting import crossviews_plot
import numpy as np
n_samples = 100
centers = [[0, 1], [0, -1]]
covariances = [np.eye(2), np.eye(2)]
GM = GaussianMixture(n_samples, centers, covariances, shuffle=True)
GM = GM.sample_views(transform='poly', n_noise=2)
# Below, we see that the first two dimensions are related by a degree 2
# polynomial while the latter two dimensions are uncorrelated.
crossviews_plot(GM.Xs_, labels=GM.y_,
title='View 1 vs. View 2 (Polynomial \
Transform + noise)', equal_axes=True)