def test_reconstruct_with_without_rowvar():
"""Tests the reconstruct works with rowvar correctly."""
np.random.seed(1)
data = np.random.normal(size=(10, 50))
pc1 = PrincipalComponents(data)
pc2 = PrincipalComponents(data.T, rowvar=True)
pc1._direct()
pc2._direct()
check_cols_match_except_sign(pc1.reconstruct(pc1.project(3)), pc2.reconstruct(pc2.project(3)).T)
def test_project_with_without_rowvar():
"""Tests the project works with rowvar correctly."""
np.random.seed(1)
data = np.random.normal(size=(10, 50))
pc1 = PrincipalComponents(data)
pc2 = PrincipalComponents(data.T, rowvar=True)
pc1._direct()
pc2._direct()
check_cols_match_except_sign(pc1.project(3), pc2.project(3).T)
check_cols_match_except_sign(pc1.project(3), pc2.project(3).T)
class SplineModelPCAGaussianMixture(GaussianMixture):
"""docstring for SplineModelGaussianMixture"""
def __init__(self, npc=None, *args, **keywords):
self._pc = PrincipalComponents(data)
self._pc._direct()
if npc is None:
npc = len(self._pc._eigval)
data = self._pc.project(npc)
super(SplineModelPCAGaussianMixture, self).__init__(ncomponent,
data,
norm=False,
*args,
**keywords)
def display(self,
covariance=None,
logalpha=None,
name=None,
extra=1,
k=3,
figures=(1, 2)):
"""Display covariances and alphas with matplotlib."""
pyplot.close('all')
super(SplineModelPCAGaussianMixture,
self).display(covariance, logalpha, name, figures[0])
figurenum = figures[1]
if PLOTTING_AVAILABLE:
print "Okay plotting"
pyplot.ioff()
pyplot.figure(figurenum)
pyplot.clf()
rows = cols = np.ceil(np.sqrt(self._ncomponent()))
if rows * cols == self._ncomponent():
rows = rows + 1
ncomp, ndim = self._means.shape
perspline = (ndim - extra) / 2
for clust in xrange(self._ncomponent()):
t = internal_knots(coef2knots(perspline))
t = np.concatenate((np.zeros(4), t, np.ones(4)))
pyplot.subplot(rows, cols, clust + 1)
means = self._pc.reconstruct(self._means[clust, :]).squeeze()
spline1, spline2 = unmix(means, k=k, extra=extra)
plot_from_spline(spline1)
plot_from_spline(spline2)
pyplot.ylim(-0.2, 1.01)
pyplot.show()
pyplot.ion()
class SplineModelPCAGaussianMixture(GaussianMixture):
"""docstring for SplineModelGaussianMixture"""
def __init__(self, npc=None, *args, **keywords):
self._pc = PrincipalComponents(data)
self._pc._direct()
if npc is None:
npc = len(self._pc._eigval)
data = self._pc.project(npc)
super(SplineModelPCAGaussianMixture, self).__init__(ncomponent, data,
norm=False, *args, **keywords)
def display(self, covariance=None, logalpha=None, name=None, extra=1,
k=3, figures=(1,2)):
"""Display covariances and alphas with matplotlib."""
pyplot.close('all')
super(SplineModelPCAGaussianMixture, self).display(covariance,
logalpha, name, figures[0])
figurenum = figures[1]
if PLOTTING_AVAILABLE:
print "Okay plotting"
pyplot.ioff()
pyplot.figure(figurenum)
pyplot.clf()
rows = cols = np.ceil(np.sqrt(self._ncomponent()))
if rows * cols == self._ncomponent():
rows = rows + 1
ncomp, ndim = self._means.shape
perspline = (ndim - extra) / 2
for clust in xrange(self._ncomponent()):
t = internal_knots(coef2knots(perspline))
t = np.concatenate((np.zeros(4),t,np.ones(4)))
pyplot.subplot(rows, cols, clust+1)
means = self._pc.reconstruct(self._means[clust, :]).squeeze()
spline1, spline2 = unmix(means,k=k,extra=extra)
plot_from_spline(spline1)
plot_from_spline(spline2)
pyplot.ylim(-0.2, 1.01)
pyplot.show()
pyplot.ion()