def categorical_pmf(model, col_idx, row_idx):
hypers, suffstats = _get_hypers_and_suffstats(model, col_idx, row_idx)
ps = np.exp([
csd.probability(i, suffstats, hypers)
for i in range(len(suffstats) - 2)
])
return ps
def _probability_component_categorical(x, model, col_idx, component_idx):
hypers = model['col_hypers'][col_idx]
if component_idx < len(model['col_suffstats'][col_idx]):
suffstats = model['col_suffstats'][col_idx][component_idx]
else:
k = int(model['col_suffstats'][col_idx][0]['k'])
suffstats = {'n': 0., 'k': k}
return csd.probability(x, suffstats, hypers)
def _probability_component_categorical(x, model, col_idx, component_idx):
hypers = model['col_hypers'][col_idx]
if component_idx < len(model['col_suffstats'][col_idx]):
suffstats = model['col_suffstats'][col_idx][component_idx]
else:
k = int(model['col_suffstats'][col_idx][0]['k'])
suffstats = {'n': 0., 'k': k}
return csd.probability(x, suffstats, hypers)
def test_probability_values_4(hypers, suffstats):
hypers['dirichlet_alpha'] = .25
suffstats['0'] = 2
suffstats['1'] = 7
suffstats['2'] = 13
suffstats['n'] = 22
csd_value = csd.probability(0, suffstats, hypers)
assert csd_value == approx(-2.31363492918062)
def categorical_pmf(model, col_idx, row_idx):
hypers, suffstats = _get_hypers_and_suffstats(model, col_idx, row_idx)
ps = np.exp([csd.probability(i, suffstats, hypers) for i in
range(len(suffstats)-2)])
return ps
def test_probability_values_3(hypers, suffstats):
hypers['dirichlet_alpha'] = 2.5
csd_value = csd.probability(0, suffstats, hypers)
assert csd_value == approx(-1.6094379124341)
def test_probability_values_2(hypers, suffstats):
csd_value = csd.probability(1, suffstats, hypers)
assert csd_value == approx(-0.95551144502744)
def test_probability_values_1(hypers, suffstats):
csd_value = csd.probability(0, suffstats, hypers)
assert csd_value == approx(-1.87180217690159)
