def setUp(self):
self.a = {chr(i + 97): list(range(10)) for i in range(5)}
self.t = Trace(**self.a)
self.mt = Trace(self.a, self.a, self.a)
self.real_mt = Trace.from_csv(FULL_PATH + r'/data/south_mvcm_5000',
multi=True)
self.real_singles = [
Trace.from_csv(FULL_PATH +
r'/data/south_mvcm_5000_{}.csv'.format(i))
for i in range(4)
]
def test_validate_names(self):
b = self.a.copy()
try:
bad_names = Trace(self.a, b, self.a, self.a)
except KeyError:
pass
def test_ordering(self):
for ch, alone in zip(self.real_mt.chains, self.real_singles):
Trace(ch)._assert_allclose(alone)
def __init__(
self,
#data parameters
Y,
X,
coordinates,
n_samples=1000,
n_jobs=1,
priors=None,
configs=None,
starting_values=None,
extra_traced_params=None,
dmetric='euclidean',
correlation_function=nexp,
verbose=False,
center=True,
rescale_dists=True):
if center:
X = verify_center(X)
X = verify_covariates(X)
N, p = X.shape
Xs = X
X = explode(X)
self.state = Hashmap(X=X, Y=Y, coordinates=coordinates)
self.traced_params = ['Betas', 'Mus', 'T', 'Phi', 'Tau2']
if extra_traced_params is not None:
self.traced_params.extend(extra_traced_params)
self.trace = Trace(**{param: [] for param in self.traced_params})
st = self.state
self.state.correlation_function = correlation_function
self.verbose = verbose
st.Y = Y
st.X = X
st.Xs = Xs
st.N = N
st.p = p
st._dmetric = dmetric
if isinstance(st._dmetric, str):
st.pwds = d.squareform(d.pdist(st.coordinates, metric=st._dmetric))
elif callable(st._dmetric):
st.pwds = st._dmetric(st.coordinates)
st.max_dist = st.pwds.max()
if rescale_dists:
st.pwds = st.pwds / st.max_dist
st._old_max = st.max_dist
st.max_dist = 1.
if configs is None:
configs = dict()
if priors is None:
priors = dict()
if starting_values is None:
starting_values = dict()
self._setup_priors(**priors)
self._setup_starting_values(**starting_values)
self._setup_configs(**configs)
self._verbose = verbose
self.cycles = 0
if n_samples > 0:
try:
self.sample(n_samples, n_jobs=n_jobs)
except (np.linalg.LinAlgError, ValueError) as e:
Warn('Encountered the following LinAlgError. '
'Model will return for debugging. \n {}'.format(e))