def get_S(self, dimensions=2, pub=None):
similarity = self.get_similarity(self.statistic)
if pub is not None:
pub.array_as_image('similarity', similarity,
caption='Similarity statistic')
plot_spectrum(pub, 'similarity', similarity)
if self.scale_score:
R = scale_score(similarity).astype('float32')
R = R / R.max()
if pub is not None:
pub.array_as_image('scale_score', R)
else:
R = similarity
D = 1 - R
D = D * np.pi / D.max()
np.fill_diagonal(D, 0)
if pub is not None:
# pub.array_as_image('D', D)
P = D * D
B = double_center(P)
# plot_spectrum(pub, 'D', D)
# plot_spectrum(pub, 'P', P)
plot_spectrum(pub, 'B', B)
S = mds(D, ndim=dimensions)
# S = inner_product_embedding(similarity, 3)
# S = S[1:3, :]
return S
def mds_test():
for n in [10, 100]:
for k in [3, 4, 5]:
P = np.random.rand(k, n)
D = euclidean_distances(P)
P2 = mds(D, ndim=k)
error = evaluate_error(P, P2)
assert_allclose(0, error, atol=1e-7)
def mds_test():
for n in [10, 100]:
for k in [3, 4, 5]:
P = np.random.rand(k, n)
D = euclidean_distances(P)
P2 = mds(D, ndim=k)
error = evaluate_error(P, P2)
assert_allclose(0, error, atol=1e-7)
def _solve(self, R):
ndim = self.params['ndim']
D0 = R.max() - R
np.testing.assert_almost_equal(D0.min(), 0)
np.testing.assert_allclose(D0[0, 0], 0)
S = mds(D0, ndim)
self.iteration(dict(S=S))
def get_embedding(self, R, ndim):
D = 1 - R
D = D * np.pi / D.max()
np.fill_diagonal(D, 0)
# if pub is not None:
# P = D * D
# B = double_center(P)
# plot_spectrum(pub, 'B', B)
S = mds(D, ndim=ndim)
return S
def get_embedding_mds(plans, D, ndim):
""" Returns a dictionary plan -> position """
plan2index = dict([(p, i) for i, p in enumerate(plans)])
p2 = mds(D, ndim)
plan2point = dict([(p, p2[:, plan2index[p]]) for p in plans])
return plan2point
def get_embedding_mds(plans, D, ndim):
""" Returns a dictionary plan -> position """
plan2index = dict([(p, i) for i, p in enumerate(plans)])
p2 = mds(D, ndim)
plan2point = dict([(p, p2[:, plan2index[p]]) for p in plans])
return plan2point