def plot_km_em_clusters(x_train_scaled, x=0, y=1, z = 0, dataset_name = "", km_clusters = 12, em_clusters=12):
kmeans = KMeans(n_clusters=km_clusters, random_state=random_state)
y_pred = kmeans.fit_predict(x_train_scaled)
plot_clusters_3d(x_train_scaled, y_pred, x, y, z, dataset_name=dataset_name, classifier = "K-means")
gm = BayesianGaussianMixture(n_components = em_clusters, random_state=random_state, reg_covar=1e-01)
y_pred = gm.fit_predict(x_train_scaled)
plot_clusters_3d(x_train_scaled, y_pred, x, y, z, dataset_name=dataset_name, classifier = "EM")
def view_gmm_graph(selected: np.array,
validation_type: str = 'default',
df: pd.DataFrame = None,
n_init: int = 5,
max_k=150) -> tuple:
scores = []
temp_df = df.copy() if df is not None else None
temp_selected = selected.toarray()
if validation_type == 'default':
for i in range(10, max_k, 10):
model = BayesianGaussianMixture(i,
covariance_type='diag',
n_init=n_init,
init_params='kmeans',
random_state=12)
scores.append(model.fit(temp_selected).score(temp_selected))
print(f'k={i}, score={scores[-1]}')
elif validation_type == 'purity':
for i in range(10, max_k, 10):
model = BayesianGaussianMixture(i,
covariance_type='diag',
n_init=n_init,
init_params='kmeans',
random_state=12)
temp_df['KMeans'] = model.fit_predict(temp_selected)
scores.append(cluster_purity(temp_df))
print(f'k={i}, score={scores[-1]}')
plt.plot(range(10, max_k, 10), scores)
plt.xlabel('Number of clusters')
plt.ylabel('Score')
plt.show()
return list(range(10, max_k, 10)), scores
def test_bayesian_mixture_fit_predict_n_init():
# Check that fit_predict is equivalent to fit.predict, when n_init > 1
X = np.random.RandomState(0).randn(50, 5)
gm = BayesianGaussianMixture(n_components=5, n_init=10, random_state=0)
y_pred1 = gm.fit_predict(X)
y_pred2 = gm.predict(X)
assert_array_equal(y_pred1, y_pred2)
def test_bayesian_mixture_fit_predict_n_init():
# Check that fit_predict is equivalent to fit.predict, when n_init > 1
X = np.random.RandomState(0).randn(1000, 5)
gm = BayesianGaussianMixture(n_components=5, n_init=10, random_state=0)
y_pred1 = gm.fit_predict(X)
y_pred2 = gm.predict(X)
assert_array_equal(y_pred1, y_pred2)
def BayesianGaussianMixture(V, **kwargs):
"""Performs clustering on *V* by using Gaussian mixture models with variational inference. The function uses :func:`sklearn.micture.GaussianMixture`. See sklearn documents
for details.
:arg V: row-normalized eigenvectors for the purpose of clustering.
:type V: :class:`numpy.ndarray`
:arg n_clusters: specifies the number of clusters.
:type n_clusters: int
"""
try:
from sklearn.mixture import BayesianGaussianMixture
except ImportError:
raise ImportError('Use of this function (BayesianGaussianMixture) requires the '
'installation of sklearn.')
n_components = kwargs.pop('n_components', None)
if n_components == None:
n_components = kwargs.pop('n_clusters',None)
if n_components == None:
n_components = 1
n_init = kwargs.pop('n_init', 1)
mixture = BayesianGaussianMixture(n_init=n_init, **kwargs).fit(V)
return mixture.fit_predict(V)
def model_dpgmm(k, x_scaled):
global labels
print("\nDirichlet Process Gaussian Mixture ... [" + str(len(x_scaled)) +
' training samples -> ' + str(k) + ' initial clusters]')
model = BayesianGaussianMixture(n_components=k, covariance_type='full')
labels = model.fit_predict(x_scaled)
print('Done! Final cluster numbers = ', len(np.unique(labels)))
do_dataframe()
def identifyTransitionsComplete(p, window_size):
joint_angles = (p['X'][:, :, 0:7])
joint_velocity = (p['X'][:, :, 7:14])
torque = (p['U'][:, :, :])
endeff_pose = (p['EX'][:, :, 0:6])
endeff_velocity = (p['EX'][:, :, 6:12])
force_feedback = (p['F'][:, :, :])
total_rollout = joint_velocity.shape[0]
traj = []
for index in range(0, total_rollout):
traj_feature = np.hstack(
(joint_angles[index, :, :], endeff_pose[index, :, :],
force_feedback[index, :, :]))
traj.append(traj_feature)
traj = np.array(traj)
traj_time = traj.shape[1]
dim = traj.shape[2]
total_size = total_rollout * traj_time
demo_data_array = np.zeros((total_size - window_size, dim * window_size))
inc = 0
for j in range(0, total_rollout):
trajC = traj[j, :, :]
for i in range(window_size, traj_time):
window = trajC[i - window_size:i, :]
demo_data_array[inc, :] = np.reshape(window,
(1, dim * window_size))
inc = inc + 1
estimator = BayesianGaussianMixture(n_components=10,
n_init=10,
max_iter=300,
weight_concentration_prior=1e-1,
init_params='random',
verbose=False)
labels = estimator.fit_predict(demo_data_array)
# print(estimator.weights_)
filtabels = smoothing(labels)
# print(labels)
inc = 0
transitions = []
for j in range(window_size, total_size):
if inc == 0 or j == window_size:
pass # self._transitions.append((i,0))
elif j == (total_size - 1):
pass # self._transitions.append((i,n-1))
elif filtabels[inc - 1] != filtabels[inc]:
transitions.append(j - window_size)
inc = inc + 1
transitions.append(0)
transitions.append(total_size - 1)
transitions.sort()
print("[TSC] Discovered Transitions (number): ", len(transitions))
return transitions
def partition_data(self,j):
dp = BayesianGaussianMixture(n_components = int(self.alpha*np.log(self.N)) ,weight_concentration_prior = self.alpha, init_params='kmeans',weight_concentration_prior_type='dirichlet_process')
Z = dp.fit_predict(self.X[self.U[j]])
le = LE()
Z = le.fit_transform(Z)
Z_count = np.bincount(Z)
assert(Z.max()+1 == Z_count.size)
self.K[j] = int(Z_count.size)
self.marginal_LL_k[j] = {k:0 for k in range(int(self.K[j])) }
return(Z,Z_count)
def cluster(points,
clouds=None,
concentration_prior=None,
K=100,
restarts=10,
seed=0):
"""
Clusters a set of data points lying in an arbitrary number of clusters.
Arguments:
data (list of lists of floats): list of data points to be clustered.
clouds (list or lists of floats, same second dimension as data): bootstrapped bins for clustering
sampleName (string): The name of the input sample.
concentration_prior (float): Tuning parameter for clustering, must be between 0 and 1. Used to determine concentration
of points in clusters -- higher favors more clusters, lower favors fewer clusters.
K (int): maximum number of clusters to infer
restarts (int): number of initializations to try for GMM
seed (int): random number generator seed for GMM
Returns:
mus (list of lists of floats): List of cluster means.
sigmas (list of 2D lists of floats): List of cluster covariances.
clusterAssignments (list of ints): The assignment of each interval to a cluster, where an entry
j at index i means the ith interval has been assigned to the
jth meta-interval.
numPoints (list of ints): Number of points assigned to each cluster
numClusters (int): The number of clusters.
"""
from sklearn.mixture import BayesianGaussianMixture
from collections import Counter
sp.log(msg="## Clustering with K={} and c={}...\n".format(
K, concentration_prior),
level="INFO")
total = list(points)
if clouds is not None:
total.extend(list(clouds))
npArray = np.array(total)
gmm = BayesianGaussianMixture(
n_components=K,
n_init=restarts,
weight_concentration_prior=concentration_prior,
max_iter=int(1e6),
random_state=seed)
targetAssignments = gmm.fit_predict(npArray)
targetAssignments = targetAssignments[:len(points)]
mus = gmm.means_
sigmas = gmm.covariances_
cntr = Counter(targetAssignments)
numPoints = [cntr[i] if i in cntr else 0 for i in range(K)]
numClusters = len(cntr)
return mus, sigmas, targetAssignments, numPoints, numClusters
def add_gmm_labels(df: pd.DataFrame,
selected: np.array,
k: int = 60,
n_init=5):
dense_selected = selected.toarray()
model = BayesianGaussianMixture(k,
covariance_type='diag',
n_init=n_init,
init_params='kmeans',
random_state=12)
labels = model.fit_predict(dense_selected)
df['GMM'] = labels
return model
def ms_VB(self):
# 直接使用 sklearn 中的 VBGMM
self.K = MAX_K
clf = BayesianGaussianMixture(n_components=MAX_K,
covariance_type="full",
max_iter=200,
random_state=0)
y = self.correct_order(clf.fit_predict(self.x), clf)
self.mus = clf.means_
print(y)
self.show_scatter(y, "VBGMM")
accuracy = np.mean(self.real_y.ravel() == y.ravel())
print(accuracy)
def relabel(dataset, n_components=4):
if not len(dataset):
return np.array([])
new_labels = -1 * np.ones(len(dataset.labels))
for l in np.unique(dataset.labels):
gmm = BayesianGaussianMixture(n_components=n_components,
weight_concentration_prior=1 /
(n_components * 2),
max_iter=200)
ttt = PCA(n_components=6).fit_transform(
dataset.waveforms[dataset.labels == l])
gmm_labels = gmm.fit_predict(ttt)
new_labels[dataset.labels == l] = gmm_labels + 1 + np.max(new_labels)
return new_labels
def identifyTransitions(traj, window_size, weight_prior, n_components):
'''
Transition detection function based on DPGMM and windowing approach
:param traj: trajectory with states, action, contact forces
:param window_size: windows size used to accumulate states
:param weight_prior, n_components: parameter used for DPGMM
:return: transition points (index) in the trajectory
'''
total_size = traj.shape[0]
dim = traj.shape[1]
demo_data_array = np.zeros((total_size - window_size, dim * window_size))
inc = 0
for i in range(window_size, total_size):
window = traj[i - window_size:i, :]
demo_data_array[inc, :] = np.reshape(window, (1, dim * window_size))
inc = inc + 1
estimator = BayesianGaussianMixture(
n_components=n_components,
n_init=10,
max_iter=300,
weight_concentration_prior=weight_prior,
init_params='random',
verbose=False)
labels = estimator.fit_predict(demo_data_array)
filtabels = smoothing(labels)
inc = 0
transitions = []
for j in range(window_size, total_size):
if inc == 0 or j == window_size:
pass # self._transitions.append((i,0))
elif j == (total_size - 1):
pass # self._transitions.append((i,n-1))
elif filtabels[inc - 1] != filtabels[inc]:
transitions.append(j - window_size)
inc = inc + 1
transitions.append(0)
transitions.append(total_size - 1)
transitions.sort()
# print("[TSC] Discovered Transitions (number): ", len(transitions))
return transitions
def execute(self, namespace):
from sklearn.mixture import GaussianMixture, BayesianGaussianMixture
from PYME.IO import MetaDataHandler
points = namespace[self.input_points]
X = np.stack([points['x'], points['y'], points['z']], axis=1)
if self.mode == 'n':
gmm = GaussianMixture(n_components=self.n,
covariance_type=self.covariance)
predictions = gmm.fit_predict(X)
elif self.mode == 'bic':
n_components = range(1, self.n + 1)
bic = np.zeros(len(n_components))
for ind in range(len(n_components)):
gmm = GaussianMixture(n_components=n_components[ind],
covariance_type=self.covariance)
gmm.fit(X)
bic[ind] = gmm.bic(X)
logger.debug('%d BIC: %f' % (n_components[ind], bic[ind]))
best = n_components[np.argmin(bic)]
if best == self.n or (self.n > 10 and best > 0.9 * self.n):
logger.warning(
'BIC optimization selected n components near n max')
gmm = GaussianMixture(n_components=best,
covariance_type=self.covariance)
predictions = gmm.fit_predict(X)
elif self.mode == 'bayesian':
bgm = BayesianGaussianMixture(n_components=self.n,
covariance_type=self.covariance)
predictions = bgm.fit_predict(X)
out = tabular.MappingFilter(points)
try:
out.mdh = MetaDataHandler.DictMDHandler(points.mdh)
except AttributeError:
pass
out.addColumn(self.label_key, predictions)
namespace[self.output_labeled] = out
def get_direction(self, X, mu):
r"""
Generate direction vectors.
Parameters
----------
X : array
Array of shape ``(nwalkers//2, ndim)`` with the walker positions of the complementary ensemble.
mu : float
The value of the scale factor ``mu``.
Returns
-------
directions : array
Array of direction vectors of shape ``(nwalkers//2, ndim)``.
"""
if not self.tune:
mu = self.mu0
n = X.shape[0]
mixture = BayesianGaussianMixture(n_components=self.n_components)
labels = mixture.fit_predict(X)
means = mixture.means_
covariances = mixture.covariances_
i, j = np.random.choice(labels, 2, replace=False)
if i != j:
directions = np.random.multivariate_normal(
means[i], covariances[i] * self.rescale_cov,
size=n) - np.random.multivariate_normal(
means[j], covariances[j] * self.rescale_cov, size=n)
tune_once = False
else:
directions = mu * np.random.multivariate_normal(
np.zeros_like(means[i]), covariances[i], size=n)
if self.tune:
tune_once = True
else:
tune_once = False
return 2.0 * directions, tune_once
def identifyTransitions(traj, window_size):
"""
Identify transition by accumulating data points using sliding window and using DP GMM to find
clusters in a single trajectory
"""
total_size = traj.shape[0]
dim = traj.shape[1]
demo_data_array = np.zeros((total_size - window_size, dim * window_size))
inc = 0
for i in range(window_size, total_size):
window = traj[i - window_size:i, :]
demo_data_array[inc, :] = np.reshape(window, (1, dim * window_size))
inc = inc + 1
estimator = BayesianGaussianMixture(n_components=5,
n_init=10,
max_iter=300,
weight_concentration_prior=0.01,
init_params='random',
verbose=False)
labels = estimator.fit_predict(demo_data_array)
# print(estimator.weights_)
filtabels = smoothing(labels)
# print(labels)
inc = 0
transitions = []
for j in range(window_size, total_size):
if inc == 0 or j == window_size:
pass # self._transitions.append((i,0))
elif j == (total_size - 1):
pass # self._transitions.append((i,n-1))
elif filtabels[inc - 1] != filtabels[inc]:
transitions.append(j - window_size)
inc = inc + 1
transitions.append(0)
transitions.append(total_size - 1)
transitions.sort()
print("[TSC] Discovered Transitions (number): ", len(transitions))
return transitions
def identifyTransitions(traj, window_size):
total_size = traj.shape[0]
dim = traj.shape[1]
demo_data_array = np.zeros((total_size - window_size, dim * window_size))
inc = 0
for i in range(window_size, total_size):
window = traj[i - window_size:i, :]
demo_data_array[inc, :] = np.reshape(window, (1, dim * window_size))
inc = inc + 1
estimator = BayesianGaussianMixture(n_components=10,
n_init=10,
max_iter=300,
weight_concentration_prior=1e-2,
init_params='random',
verbose=False)
labels = estimator.fit_predict(demo_data_array)
# print(estimator.weights_)
filtabels = smoothing(labels)
# print(labels)
inc = 0
transitions = []
for j in range(window_size, total_size):
if inc == 0 or j == window_size:
pass # self._transitions.append((i,0))
elif j == (total_size - 1):
pass # self._transitions.append((i,n-1))
elif filtabels[inc - 1] != filtabels[inc]:
transitions.append(j - window_size)
inc = inc + 1
transitions.append(0)
transitions.append(total_size - 1)
transitions.sort()
# print("[TSC] Discovered Transitions (time): ", transitions)
return transitions
def local_gmm(global_label, levels, n=10, image=None):
assert (
type(image) is np.ndarray
and image.dtype == np.uint8
and len(image.shape) == 2
or image is None
), "The input image has to be a uint8 2D numpy array, or omitted."
assert (
type(global_label) is np.ndarray
and global_label.dtype == np.uint8
and len(global_label.shape) == 2
), "The input global_label has to be a uint8 2D numpy array."
assert len(levels) == np.max(
global_label
), "The nubmer of levels should match that of the global labels."
assert type(n) is int
local_label = np.zeros(global_label.shape, dtype=np.uint8)
blob_levels = []
for i in range(len(levels)):
lvl_ind = (global_label == i + 1).nonzero()
data = np.transpose(lvl_ind)
gmm = BayesianGaussianMixture(n_components=min(n, len(data)), random_state=123)
prediction = gmm.fit_predict(data)
for j in np.unique(prediction):
blob_levels.append(levels[i])
blob_pts = prediction == j
local_label[lvl_ind[0][blob_pts], lvl_ind[1][blob_pts]] = len(blob_levels)
label_resized = (
local_label
if image.shape == local_label.shape
else cv2.resize(local_label, image.shape[::-1], interpolation=cv2.INTER_NEAREST)
)
image_labeled = (
None
if image is None
else img_as_ubyte(label2rgb(label_resized, image, bg_label=0))
)
return image_labeled, local_label, blob_levels
def cluster_vbgm(aligned_maps):
# sample_by_features = np.vstack([xmap.flatten() for xmap in aligned_maps])
embedding = embed(aligned_maps)
clusterer = BayesianGaussianMixture(n_components=10)
return clusterer.fit_predict(embedding)
n_features=2,
center_box=[-5, 5],
centers=nb_centers,
random_state=1000)
# Train the model with concentration 1000 and 0.1
for c in (1000.0, 0.1):
gm = BayesianGaussianMixture(n_components=5,
weight_concentration_prior=c,
max_iter=10000,
random_state=1000)
gm.fit(X)
print('Weights: {}'.format(gm.weights_))
Y_pred = gm.fit_predict(X)
print((Y_pred == 0).sum())
print((Y_pred == 1).sum())
print((Y_pred == 2).sum())
print((Y_pred == 3).sum())
print((Y_pred == 4).sum())
# Compute the parameters of the Gaussian mixture
m1 = gm.means_[0]
m2 = gm.means_[1]
m3 = gm.means_[2]
m4 = gm.means_[3]
m5 = gm.means_[4]
c1 = gm.covariances_[0]
# for i in range(len(nc_array)):
# dp = GaussianMixture(n_components=nc_array[i], covariance_type='full', max_iter=10000, verbose=0)
# dp.fit(label_data)
#
# dpgmm_list.append(dp)
# log_acc[i] = dp.lower_bound_ - nc_array[i]/rho
#
# dpgmm = dpgmm_list[np.argmax(log_acc)]
# print(len(dpgmm.covariances_), log_acc)
dpgmm = BayesianGaussianMixture(n_components=n_components, covariance_type='full', max_iter=50000,
weight_concentration_prior_type='dirichlet_process', mean_precision_prior=.8,
weight_concentration_prior=gamma, n_init=1, init_params='random', reg_covar=1e-6)
# gmm_labels = dpgmm.predict(label_data)
gmm_labels = dpgmm.fit_predict(label_data)
# print('Variational', max(gmm_labels)+1)
for k, cov in enumerate(dpgmm.covariances_):
c1, c2 = np.diag(cov)
if c1/c2 < cov_ratio and c2/c1 < cov_ratio:
em_data.extend(label_data[gmm_labels == k])
# import pdb; pdb.set_trace()
if np.mod(m, 10) == 0:
print('cluster {} out of {}'.format(m, len(arg_labels)))
# plot_results(label_data, gmm_labels, dpgmm.means_, dpgmm.covariances_)
# plt.show()
def model_dpgmm(k, x_scaled):
global labels
model = BayesianGaussianMixture(n_components=k, covariance_type='full')
labels = model.fit_predict(x_scaled)
X_pen_scaled = X_pen_scaled.reshape(-1, 1)
X_train_pen, X_test_pen, y_train_pen, y_test_pen = train_test_split(
X_pen_scaled, ypen, test_size=0.20)
pen_classifier.fit(X_train_pen, y_train_pen)
pen_pred = pen_classifier.predict(X_test_pen)
pen_error_kmean.append(1 - metrics.accuracy_score(pen_pred, y_test_pen))
#===========================================================
#===========================EM=============================
from sklearn.decomposition import FastICA
for i in range(1, 31):
X_pen_scaled = pen_scaler.fit_transform(Xpen)
pen_bgm = BayesianGaussianMixture(n_components=i)
X_pen_scaled = pen_bgm.fit_predict(X_pen_scaled)
X_pen_scaled = X_pen_scaled.reshape(-1, 1)
X_train_pen, X_test_pen, y_train_pen, y_test_pen = train_test_split(
X_pen_scaled, ypen, test_size=0.20)
pen_classifier.fit(X_train_pen, y_train_pen)
pen_pred = pen_classifier.predict(X_test_pen)
pen_error_em.append(1 - metrics.accuracy_score(pen_pred, y_test_pen))
#===========================================================
plt.figure(figsize=(12, 6))
plt.plot(range(1, 31),
pen_error,
label='No Clustering',
color='red',
row_ix = np.where(yhat == cluster)
# create scatter of these samples
plt.scatter(X[row_ix, 0], X[row_ix, 1])
# show the plot
plt.show()
# advanced
#================================================
ddgmm = BayesianGaussianMixture(
n_components=2,
covariance_type='full',
weight_concentration_prior=100,
weight_concentration_prior_type="dirichlet_distribution",
max_iter=100,
random_state=1337).fit(X)
yhat = ddgmm.fit_predict(X)
# retrieve unique clusters
clusters = np.unique(yhat)
# create scatter plot for samples from each cluster
for cluster in clusters:
# get row indexes for samples with this cluster
row_ix = np.where(yhat == cluster)
# create scatter of these samples
plt.scatter(X[row_ix, 0], X[row_ix, 1])
# show the plot
plt.show()
dpgmm = BayesianGaussianMixture(
n_components=2,
covariance_type='full',
weight_concentration_prior=100,
print(test_mean[np.argmax(test_mean)])
print(train_mean[np.argmax(test_mean)])
mlp_learner = MLPClassifier(hidden_layer_sizes=(100,),activation='relu',solver='sgd', learning_rate = 'adaptive', learning_rate_init = 0.07)
train_sizes, train_scores, test_scores = learning_curve(mlp_learner, x_train_scaled_km, y_train, cv=5, n_jobs=-1, train_sizes=np.linspace(0.01, 1.0, 100))
train_mean = np.mean(train_scores, axis=1)
test_mean = np.mean(test_scores, axis=1)
plot_data(train_sizes, test_mean, title="Neural Network Learning Curve - Dataset 1 + Clustering Result", x_label="Training Size", y_label="Accuracy Score", color="orange", label='CV (+K-means Result)', linestyle='dashed')
plot_data(train_sizes, train_mean, title="Neural Network Learning Curve - Dataset 1 + Clustering Result", x_label="Training Size", y_label="Accuracy Score", color="orange", label='Training (+K-means Result')
print(train_sizes)
print((train_sizes[np.argmax(test_mean)]))
print(test_mean[np.argmax(test_mean)])
print(train_mean[np.argmax(test_mean)])
gm = BayesianGaussianMixture(n_components = 14, random_state=random_state, reg_covar=1e-01)
y_pred = gm.fit_predict(x_projected_pca)
x_train_scaled_em = np.column_stack((x_train_scaled,y_pred))
mlp_learner = MLPClassifier(hidden_layer_sizes=(100,),activation='relu',solver='sgd', learning_rate = 'adaptive', learning_rate_init = 0.07)
train_sizes, train_scores, test_scores = learning_curve(mlp_learner, x_train_scaled_em, y_train, cv=5, n_jobs=-1, train_sizes=np.linspace(0.01, 1.0, 100))
train_mean = np.mean(train_scores, axis=1)
test_mean = np.mean(test_scores, axis=1)
plot_data(train_sizes, test_mean, title="Neural Network Learning Curve - Dataset 1 + Clustering Result", x_label="Training Size", y_label="Accuracy Score", color="blue", label='CV (+EM Result)', linestyle='dashed')
plot_data(train_sizes, train_mean, title="Neural Network Learning Curve - Dataset 1 + Clustering Result", x_label="Training Size", y_label="Accuracy Score", color="blue", label='Training (+EM Result)')
print(train_sizes)
print((train_sizes[np.argmax(test_mean)]))
print(test_mean[np.argmax(test_mean)])
print(train_mean[np.argmax(test_mean)])
plt.savefig('Dataset 1 + Clustering NN learning curve.png')
scl = [0, "rgb(150,0,90)"], [0.125, "rgb(0, 0, 200)"], [0.25, "rgb(0, 25, 255)"], \
[0.375, "rgb(0, 152, 255)"], [0.5, "rgb(44, 255, 150)"], [0.625, "rgb(151, 255, 0)"], \
[0.75, "rgb(255, 234, 0)"], [0.875, "rgb(255, 111, 0)"], [1, "rgb(255, 0, 0)"]
if __name__ == '__main__':
scoords = SitesCoords()
sites_i = 31265
sites_f = 12100
nc = 50
mutual = True
lsites = scoords.get_direct_neighbors(sites_i, 0.35)
# lsites = range(sites_i, sites_f)
lclust = compute_clusterings(lsites, nc, mutual=mutual)
mdist = compute_distance_matrix(lclust, mutual=mutual)
#plot_md_scaling(mdist)
tdata = md_scaling(mdist)
#cs = adjust_nc(tdata)
#kmeans = KMeans(n_clusters=cs)
#labels = kmeans.fit_predict(tdata)
gmm = BayesianGaussianMixture(n_components=10,
covariance_type='full',
max_iter=1000,
n_init=10,
tol=0.00001)
labels = gmm.fit_predict(tdata)
create_plot(data_plot(lsites, labels), str(sites_i))
#============================1. Uncomment this section for use with plots A, B, C, D =======================================================
for i in range(20):
from sklearn.mixture import BayesianGaussianMixture # Chose BayesianGaussianMixture because it received better accuracy scores than just GaussianMixture
bgm_sat = BayesianGaussianMixture(
n_components=7,
covariance_type='tied',
weight_concentration_prior=params[i],
max_iter=500) # 7 categories, domain knowledge
bgm_pen = BayesianGaussianMixture(
n_components=10,
covariance_type='full',
weight_concentration_prior=params[i],
max_iter=500) # 10 categories, domain knowledge
start_time = time.time()
sat_labels_pred = bgm_sat.fit_predict(X_train_sat_og)
#=======2. Use only for Plot C==================================
# sat_labels_train = bgm_sat.fit_predict(X_train_sat_og)
# sat_labels_test = bgm_sat.predict(X_test_sat_og)
#=======================================================
end_time = time.time()
sat_time = end_time - start_time
start_time = time.time()
pen_labels_pred = bgm_pen.fit_predict(X_train_pen_og)
#==========3. Use only for making Plot C======================
# pen_labels_train = bgm_sat.fit_predict(X_train_pen_og)
# pen_labels_test = bgm_sat.predict(X_test_pen_og)
#=====================================================
end_time = time.time()
pen_time = end_time - start_time
#=====================================================================================================================================
def execute(self, namespace):
from sklearn.mixture import GaussianMixture, BayesianGaussianMixture
from PYME.IO import MetaDataHandler
points = namespace[self.input_points]
X = np.stack([points['x'], points['y'], points['z']], axis=1)
if self.mode == 'n':
gmm = GaussianMixture(n_components=self.n,
covariance_type=self.covariance,
max_iter=self.max_iter,
init_params=self.init_params)
predictions = gmm.fit_predict(X) + 1 # PYME labeling scheme
log_prob = gmm.score_samples(X)
if not gmm.converged_:
logger.error('GMM fitting did not converge')
predictions = np.zeros(len(points), int)
log_prob = -np.inf * np.ones(len(points))
elif self.mode == 'bic':
n_components = range(1, self.n + 1)
bic = np.zeros(len(n_components))
for ind in range(len(n_components)):
gmm = GaussianMixture(n_components=n_components[ind],
covariance_type=self.covariance,
max_iter=self.max_iter,
init_params=self.init_params)
gmm.fit(X)
bic[ind] = gmm.bic(X)
logger.debug('%d BIC: %f' % (n_components[ind], bic[ind]))
best = n_components[np.argmin(bic)]
if best == self.n or (self.n > 10 and best > 0.9 * self.n):
logger.warning(
'BIC optimization selected n components near n max')
gmm = GaussianMixture(n_components=best,
covariance_type=self.covariance,
max_iter=self.max_iter,
init_params=self.init_params)
predictions = gmm.fit_predict(X) + 1 # PYME labeling scheme
log_prob = gmm.score_samples(X)
if not gmm.converged_:
logger.error('GMM fitting did not converge')
predictions = np.zeros(len(points), int)
log_prob = -np.inf * np.ones(len(points))
elif self.mode == 'bayesian':
bgm = BayesianGaussianMixture(n_components=self.n,
covariance_type=self.covariance,
max_iter=self.max_iter,
init_params=self.init_params)
predictions = bgm.fit_predict(X) + 1 # PYME labeling scheme
log_prob = bgm.score_samples(X)
if not bgm.converged_:
logger.error('GMM fitting did not converge')
predictions = np.zeros(len(points), int)
log_prob = -np.inf * np.ones(len(points))
out = tabular.MappingFilter(points)
try:
out.mdh = MetaDataHandler.DictMDHandler(points.mdh)
except AttributeError:
pass
out.addColumn(self.label_key, predictions)
out.addColumn(self.label_key + '_log_prob', log_prob)
avg_log_prob = np.empty_like(log_prob)
for label in np.unique(predictions):
mask = label == predictions
avg_log_prob[mask] = np.mean(log_prob[mask])
out.addColumn(self.label_key + '_avg_log_prob', avg_log_prob)
namespace[self.output_labeled] = out
n_components=args.nclusters,
covariance_type='diag',
max_iter=1000,
weight_concentration_prior_type='dirichlet_process')
dimred = TSNE(n_components=2)
fig2, ax2 = plt.subplots(1, 1)
cmap = iter([plt.cm.tab20(x) for x in range(0, 20)])
with torch.no_grad():
diter = iter(train_loader)
y, lab = diter.next()
mu, lvar = model.encode(y.view(-1, n_genes))
y2 = model.reparam(mu, lvar)
clustering.fit(y2.numpy())
idx = clustering.fit_predict(y2.numpy())
# idx = km.fit_predict(y2.numpy())
dr = dimred.fit_transform(y2.numpy())
mx, mn = np.max(dr), np.min(dr)
ax2.set_xlim([mn, mx])
for ii in np.unique(idx):
clr = np.array(next(cmap)).reshape(1, -1)
ax2.scatter(dr[idx == ii, 0], dr[idx == ii, 1], c=clr, **params)
fig2.tight_layout()
fig2.savefig(osp.join(gdir, ''.join([tag, '_tsne.png'])))
inlatent = pd.DataFrame(y2.numpy(),
index=lab,
columns=[str(x) for x in range(args.latent_dim)])
def train(data:np.ndarray,
obs_len:int,
filter_name:str,
model_dir:str,
result_dir:str,
save_model:bool=True)->NoReturn:
print('[Bayesian Gaussian Mixture Clustering][train] creating model...')
bgm = BayesianGaussianMixture(n_components=3,
covariance_type="full",
max_iter=1000,
tol=1e-5,
n_init=10,
random_state=7,
weight_concentration_prior_type='dirichlet_process',
init_params="kmeans")
print('[Bayesian Gaussian Mixture Clustering][train] training...')
_y = bgm.fit_predict(X=data)
_y = np.expand_dims(_y, axis=1)
print(f'[Bayesian Gaussian Mixture Clustering][train] converged?:{bgm.converged_}')
print('[Bayesian Gaussian Mixture Clustering][train] params (center and covariance):')
for i, m, c, w in zip(range(1, 4), bgm.means_, bgm.covariances_, bgm.weights_):
print(f'\tc_{i}-> mean: {m}')
print(f'\t\tcov: {c}')
print(f'\t\tweight: {w}')
print('[Bayesian Gaussian Mixture Clustering][train] results:')
_c, _l = np.unique(_y, return_counts=True)
for i, c in zip(_c,_l):
print (f'\tc_{i}: {c}')
if save_model:
model_file=f'bgm_{obs_len}s_{filter_name}.pkl'
print (f'[Bayesian Gaussian Mixture Clustering][train] saving model ({model_file})...')
with open(os.path.join(model_dir, model_file), 'wb') as f:
pickle.dump(bgm, f)
result_file = f'results_bgm_train_{obs_len}s_{filter_name}.csv'
print (f'[Bayesian Gaussian Mixture Clustering][train] saving results ({result_file})...')
labels = ['mean_velocity',
'mean_acceleration',
'mean_deceleration',
'std_lateral_jerk',
'driving_style']
result = np.concatenate((data, _y), axis=1)
df = pd.DataFrame(data=result, columns=labels)
df.to_csv(os.path.join(result_dir,result_file))
result_file = result_file.replace('results', 'params').replace('csv', 'json')
print (f'[Bayesian Gaussian Mixture Clustering][train] saving results ({result_file})...')
_d = {}
_d['means'] = bgm.means_.tolist()
_d['covariances'] = bgm.covariances_.tolist()
_d['weights'] = bgm.weights_.tolist()
with open(os.path.join(result_dir, result_file), 'w') as f:
json.dump(_d, f)
def km_em(x_train_scaled, dataset_name="", true_vals = y_train, reg_covar = 1e-01):
distortions = []
sil = []
n = 22
# v_measure = []
homogeneity = []
completeness = []
mutual_info = []
adj_rand_score = []
sil = []
kmeans_times = []
homogeneity_em = []
completeness_em = []
mutual_info_em = []
adj_rand_score_em = []
sil_em = []
em_times = []
em_likelihood = []
for i in range(2,n+1):
# print(i)
start_time = time.time()
kmeans = KMeans(n_clusters=i, random_state=random_state)
kmeans.fit(x_train_scaled)
distortions.append(kmeans.inertia_)
y_pred = kmeans.predict(x_train_scaled)
kmeans_times.append(time.time()-start_time)
homogeneity.append(homogeneity_score(true_vals, y_pred.tolist()))
completeness.append(completeness_score(true_vals, y_pred.tolist()))
mutual_info.append(adjusted_mutual_info_score(true_vals, y_pred.tolist()))
adj_rand_score.append(adjusted_rand_score(true_vals, y_pred.tolist()))
sil.append(silhouette_score(x_train_scaled, kmeans.labels_, metric='euclidean'))
start_time = time.time()
gm = BayesianGaussianMixture(n_components = i, random_state=random_state, reg_covar=reg_covar)
y_pred = gm.fit_predict(x_train_scaled)
em_times.append(time.time()-start_time)
homogeneity_em.append(homogeneity_score(true_vals, y_pred.tolist()))
completeness_em.append(completeness_score(true_vals, y_pred.tolist()))
mutual_info_em.append(adjusted_mutual_info_score(true_vals, y_pred.tolist()))
adj_rand_score_em.append(adjusted_rand_score(true_vals, y_pred.tolist()))
if len(set(y_pred))>1:
sil_em.append(silhouette_score(x_train_scaled, y_pred, metric='euclidean'))
else:
sil_em.append(1)
em_likelihood.append(gm.score(x_train_scaled))
# plot
plt.plot(range(2, n+1), distortions, marker='o')
plt.title("K-means Elbow ("+(str(dataset_name))+")")
plt.xlabel('Number of clusters')
plt.ylabel('Sum of Squared Distances')
plt.savefig((str(dataset_name))+' km elbow.png')
plt.show()
plt.plot(range(2, n+1), sil, marker='o')
plt.title('K-means Silhouette Scores ('+(str(dataset_name))+')')
plt.xlabel('Number of clusters')
plt.ylabel('Silhouette Score')
plt.savefig((str(dataset_name))+' km silho.png')
plt.show()
plt.plot(range(2, n+1), em_likelihood, marker='o')
plt.title('EM likelihood ('+(str(dataset_name))+')')
plt.xlabel('Number of clusters')
plt.ylabel('Likelihood')
plt.savefig((str(dataset_name))+' em likelihood.png')
plt.show()
plt.plot(range(2, n+1), sil_em, marker='o')
plt.title('EM Silhouette Scores ('+(str(dataset_name))+')')
plt.xlabel('Number of clusters')
plt.ylabel('Silhouette Score')
plt.savefig((str(dataset_name))+' em silho.png')
plt.show()
plt.close()
plot_data(list(range(1, n)), homogeneity, title="Performance Evaluation k-means ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="blue", label='Homogeneity')
plot_data(list(range(1, n)), completeness, title="Performance Evaluation k-means ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="orange", label='Completeness')
plot_data(list(range(1, n)), mutual_info, title="Performance Evaluation k-means ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="red", label='Adgusted Mutual Info')
plot_data(list(range(1, n)), adj_rand_score, title="Performance Evaluation k-means ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="green", label='Adjusted random index')
# plot_data(list(range(1, n)), v_measure, title="Performance Evaluation k-means", x_label="Number of Clusters", y_label="Score", color="brown", label='V-measure')
plt.savefig((str(dataset_name))+' km perfo.png')
plt.show()
plt.close()
plot_data(list(range(1, n)), homogeneity_em, title="Performance Evaluation EM ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="blue", label='Homogeneity')
plot_data(list(range(1, n)), completeness_em, title="Performance Evaluation EM ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="orange", label='Completeness')
plot_data(list(range(1, n)), mutual_info_em, title="Performance Evaluation EM ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="red", label='Adgusted Mutual Info')
plot_data(list(range(1, n)), adj_rand_score_em, title="Performance Evaluation EM ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Score", color="green", label='Adjusted random index')
# plot_data(list(range(1, n)), v_measure, title="Performance Evaluation EM", x_label="Number of Clusters", y_label="Score", color="brown", label='V-measure')
plt.savefig((str(dataset_name))+' em perfo.png')
plt.show()
plt.close()
plot_data(list(range(1, n)), kmeans_times, title="k-means/EM Running Time ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Time", color="red", label='k-means')
plot_data(list(range(1, n)), em_times, title="k-means/EM Running Time ("+(str(dataset_name))+")", x_label="Number of Clusters", y_label="Time", color="blue", label='EM')
plt.savefig((str(dataset_name))+' km-em time.png')
plt.show()
print('kmeans_times')
print(kmeans_times)
print('em_times')
print(em_times)
return {'sil': sil, 'kmeans_times':kmeans_times, 'em_times':em_times, 'homogeneity':homogeneity, 'completeness':completeness, 'mutual_info':mutual_info, 'adj_rand_score':adj_rand_score, 'homogeneity_em':homogeneity_em, 'completeness_em':completeness_em, 'mutual_info_em':mutual_info_em, 'adj_rand_score_em':adj_rand_score_em}