def setUp(self):
np.random.seed(0)
self.mean = np.array([[0., -2.], [0., 0.], [0., 2.]])
self.cov = np.array([[2., 0.], [0., .2]])
n_obs = 2000
data = np.vstack([np.array(rmvnorm(self.mean[k, :], self.cov, n_obs))
for k in range(self.mean.shape[0])])
self.model = FlowVB(data, 6, thresh=1e-6, init_method='random')
def setUp(self):
np.random.seed(0)
self.mean = np.array([[0., -2.], [0., 0.], [0., 2.]])
self.cov = np.array([[2., 0.], [0., .2]])
n_obs = 2000
data = np.vstack([np.array(rmvnorm(self.mean[k, :], self.cov, n_obs))
for k in range(self.mean.shape[0])])
initialiser = RandomInitialiser()
init_params = initialiser.initialise_parameters(data, 6)
options = Options(init_params)
self.model = FlowVBAnalysis(data, options)
def varsim(coefs, intercept, sig_u, steps=100, initvalues=None):
"""
Simulate simple VAR(p) process with known coefficients, intercept, white
noise covariance, etc.
"""
from numpy.random import multivariate_normal as rmvnorm
p, k, k = coefs.shape
ugen = rmvnorm(np.zeros(len(sig_u)), sig_u, steps)
result = np.zeros((steps, k))
result[p:] = intercept + ugen[p:]
# add in AR terms
for t in xrange(p, steps):
ygen = result[t]
for j in xrange(p):
ygen += np.dot(coefs[j], result[t - j - 1])
return result
def varsim(coefs, intercept, sig_u, steps=100, initvalues=None, seed=None):
"""
Simulate simple VAR(p) process with known coefficients, intercept, white
noise covariance, etc.
"""
if seed is not None:
np.random.seed(seed=seed)
from numpy.random import multivariate_normal as rmvnorm
p, k, k = coefs.shape
ugen = rmvnorm(np.zeros(len(sig_u)), sig_u, steps)
result = np.zeros((steps, k))
result[p:] = intercept + ugen[p:]
# add in AR terms
for t in xrange(p, steps):
ygen = result[t]
for j in xrange(p):
ygen += np.dot(coefs[j], result[t-j-1])
return result
import numpy as np
from numpy.random import multivariate_normal as rmvnorm
from flowvb import FlowVBAnalysis
from flowvb.core.flow_vb import Options
from flowvb.initialize import D2Initialiser
'''
Demo using synthetic data with three wide but short clusters
'''
# np.random.seed(0)
mean = np.array([[0., -2.], [0., 0.], [0., 2.]])
cov = np.array([[2., 0.], [0., .2]])
n_obs = 2000
data = np.vstack([np.array(rmvnorm(mean[k, :], cov, n_obs))
for k in range(mean.shape[0])])
num_comp_init = 6
init_params = D2Initialiser().initialise_parameters(data, num_comp_init)
options = Options(init_params)
options.plot_monitor = True
model = FlowVBAnalysis(data, options)
import numpy as np
from numpy.random import multivariate_normal as rmvnorm
from flowvb import FlowVBAnalysis
from flowvb.core.flow_vb import Options
from flowvb.initialize import D2Initialiser
'''
Demo using synthetic data with a large and a small cluster
'''
np.random.seed(0)
mean = np.array([[0., -2.], [3., 3.]])
cov = np.array([[[2., 0.], [0., .2]], [[.2, 0], [0, .2]]])
n_obs = [2000, 100]
data = np.vstack([np.array(rmvnorm(mean[k, :], cov[k, :, :], n_obs[k]))
for k in range(mean.shape[0])])
num_comp_init = 6
init_params = D2Initialiser().initialise_parameters(data, num_comp_init)
options = Options(init_params)
options.plot_monitor = True
model = FlowVBAnalysis(data, options)
