def test_torgerson(self):
data = self.ionosphere[::5]
dis = Euclidean(data)
e1 = torgerson(dis, eigen_solver="auto")
e2 = torgerson(dis, eigen_solver="lapack")
e3 = torgerson(dis, eigen_solver="arpack")
np.testing.assert_almost_equal(np.abs(e1), np.abs(e2))
np.testing.assert_almost_equal(np.abs(e2), np.abs(e3))
with self.assertRaises(ValueError):
torgerson(dis, eigen_solver="madness")
def test_torgerson(self):
data = self.ionosphere[::5]
dis = Euclidean(data)
e1 = torgerson(dis, eigen_solver="auto")
e2 = torgerson(dis, eigen_solver="lapack")
e3 = torgerson(dis, eigen_solver="arpack")
np.testing.assert_almost_equal(np.abs(e1), np.abs(e2))
np.testing.assert_almost_equal(np.abs(e2), np.abs(e3))
with self.assertRaises(ValueError):
torgerson(dis, eigen_solver="madness")
def __invalidate_embedding(self, initialization=PCA):
def jitter_coord(part):
span = np.max(part) - np.min(part)
part += np.random.uniform(-span / 20, span / 20, len(part))
# reset/invalidate the MDS embedding, to the default initialization
# (Random or PCA), restarting the optimization if necessary.
state = self.__state
if self.__update_loop is not None:
self.__set_update_loop(None)
if self.effective_matrix is None:
self.graph.reset_graph()
return
X = self.effective_matrix
if initialization == OWMDS.PCA:
self.embedding = torgerson(X)
elif initialization == OWMDS.Random:
self.embedding = np.random.rand(len(X), 2)
else:
jitter_coord(self.embedding[:, 0])
jitter_coord(self.embedding[:, 1])
self.setup_plot()
# restart the optimization if it was interrupted.
if state == OWMDS.Running:
self.__start()
def __invalidate_embedding(self, initialization=PCA):
def jitter_coord(part):
span = np.max(part) - np.min(part)
part += np.random.uniform(-span / 20, span / 20, len(part))
# reset/invalidate the MDS embedding, to the default initialization
# (Random or PCA), restarting the optimization if necessary.
state = self.__state
if self.__update_loop is not None:
self.__set_update_loop(None)
if self.effective_matrix is None:
self.graph.reset_graph()
return
X = self.effective_matrix
if initialization == OWMDS.PCA:
self.embedding = torgerson(X)
elif initialization == OWMDS.Random:
self.embedding = np.random.rand(len(X), 2)
else:
jitter_coord(self.embedding[:, 0])
jitter_coord(self.embedding[:, 1])
self.setup_plot()
# restart the optimization if it was interrupted.
if state == OWMDS.Running:
self.__start()
def __invalidate_embedding(self):
# reset/invalidate the MDS embedding, to the default initialization
# (Random or PCA), restarting the optimization if necessary.
if self.embedding is None:
return
state = self.__state
if self.__update_loop is not None:
self.__set_update_loop(None)
X = self.effective_matrix
if self.initialization == OWMDS.PCA:
self.embedding = torgerson(X)
else:
self.embedding = np.random.rand(len(X), 2)
self._update_plot()
# restart the optimization if it was interrupted.
if state == OWMDS.Running:
self.__start()
def __invalidate_embedding(self):
# reset/invalidate the MDS embedding, to the default initialization
# (Random or PCA), restarting the optimization if necessary.
if self.embedding is None:
return
state = self.__state
if self.__update_loop is not None:
self.__set_update_loop(None)
X = self.effective_matrix
if self.initialization == OWMDS.PCA:
self.embedding = torgerson(X)
else:
self.embedding = np.random.rand(len(X), 2)
self._update_plot()
# restart the optimization if it was interrupted.
if state == OWMDS.Running:
self.__start()
def setUpClass(cls):
cls.data = Table("iris")
cls.distances = Euclidean(cls.data)
cls.init = torgerson(cls.distances)
cls.args = (cls.distances, 300, 25, 0, cls.init)
