def gibbs_beta_bernoulli(relation_class, rng=None):
"""
Test with real beta-bernoulli data. Should find the two class x two class
grouping
"""
T1_N = 40
T2_N = 40
np.random.seed(0)
data = np.zeros((T1_N, T2_N) , dtype=np.bool)
data.shape = T1_N, T2_N
data[:10, :10] = True
data[10:, 10:] = True
# add noise
NOISE = 0.1
for i in range(T1_N):
for j in range(T1_N):
if np.random.rand() < NOISE:
data[i, j] = not data[i, j]
data = data[np.random.permutation(T1_N)]
data = data[:, np.random.permutation(T2_N)]
m = models.BetaBernoulli()
r = relation_class([('T1', T1_N), ('T2', T2_N)],
data,m)
hps = m.create_hps()
r.set_hps(hps)
tf_1 = model.DomainInterface(T1_N, {'r' : ('T1', r)})
tf_1.set_hps({'alpha' : 1.0})
tf_2 = model.DomainInterface(T2_N, {'r': ('T2', r)})
tf_2.set_hps({'alpha': 1.0})
### All one group for everyone
t1_g1 = tf_1.create_group(rng)
for i in range(T1_N):
tf_1.add_entity_to_group(t1_g1, i)
t2_g1 = tf_2.create_group(rng)
for i in range(T2_N):
tf_2.add_entity_to_group(t2_g1, i)
ITERS = 10
for i in range(ITERS):
gibbs.gibbs_sample_type(tf_1, rng)
gibbs.gibbs_sample_type(tf_2, rng)
print tf_1.get_assignments()
print tf_2.get_assignments()
assert_equal(len(np.unique(tf_1.get_assignments())), 2)
assert_equal(len(np.unique(tf_2.get_assignments())), 2)
def gibbs_beta_bernoulli(relation_class, rng=None):
"""
Test with real beta-bernoulli data. Should find the two class x two class
grouping
"""
T1_N = 40
T2_N = 40
np.random.seed(0)
data = np.zeros((T1_N, T2_N), dtype=np.bool)
data.shape = T1_N, T2_N
data[:10, :10] = True
data[10:, 10:] = True
# add noise
NOISE = 0.1
for i in range(T1_N):
for j in range(T1_N):
if np.random.rand() < NOISE:
data[i, j] = not data[i, j]
data = data[np.random.permutation(T1_N)]
data = data[:, np.random.permutation(T2_N)]
m = models.BetaBernoulli()
r = relation_class([('T1', T1_N), ('T2', T2_N)], data, m)
hps = m.create_hps()
r.set_hps(hps)
tf_1 = model.DomainInterface(T1_N, {'r': ('T1', r)})
tf_1.set_hps({'alpha': 1.0})
tf_2 = model.DomainInterface(T2_N, {'r': ('T2', r)})
tf_2.set_hps({'alpha': 1.0})
### All one group for everyone
t1_g1 = tf_1.create_group(rng)
for i in range(T1_N):
tf_1.add_entity_to_group(t1_g1, i)
t2_g1 = tf_2.create_group(rng)
for i in range(T2_N):
tf_2.add_entity_to_group(t2_g1, i)
ITERS = 10
for i in range(ITERS):
gibbs.gibbs_sample_type(tf_1, rng)
gibbs.gibbs_sample_type(tf_2, rng)
print tf_1.get_assignments()
print tf_2.get_assignments()
assert_equal(len(np.unique(tf_1.get_assignments())), 2)
assert_equal(len(np.unique(tf_2.get_assignments())), 2)
def test_gibbs_simple(rng=None):
"""
Test the gibbs sampling code by treating
it as a simple mixture model with the negative variance model
Max score == all T1 rows in own groups
"""
T1_N = 10
T2_N = 20
data = np.arange(T1_N * T2_N)
data.shape = T1_N, T2_N
m = models.NegVarModel()
r = relation.Relation([('T1', T1_N), ('T2', T2_N)],
data,m)
hps = m.create_hps()
hps['offset'] = 0.3
r.set_hps(hps)
tf_1 = model.DomainInterface(T1_N, {'r': ('T1', r)})
tf_1.set_hps({'alpha' : 1.0})
tf_2 = model.DomainInterface(T2_N, {'r': ('T2', r)})
tf_2.set_hps({'alpha' : 1.0})
### All one group for everyone
t1_g1 = tf_1.create_group(rng)
for i in range(T1_N):
tf_1.add_entity_to_group(t1_g1, i)
t2_g1 = tf_2.create_group(rng)
for i in range(T2_N):
tf_2.add_entity_to_group(t2_g1, i)
ITERS = 10
for i in range(ITERS):
gibbs.gibbs_sample_type(tf_1, rng)
assert_equal(len(np.unique(tf_1.get_assignments())), 10)
def test_gibbs_simple(rng=None):
"""
Test the gibbs sampling code by treating
it as a simple mixture model with the negative variance model
Max score == all T1 rows in own groups
"""
T1_N = 10
T2_N = 20
data = np.arange(T1_N * T2_N)
data.shape = T1_N, T2_N
m = models.NegVarModel()
r = relation.Relation([('T1', T1_N), ('T2', T2_N)], data, m)
hps = m.create_hps()
hps['offset'] = 0.3
r.set_hps(hps)
tf_1 = model.DomainInterface(T1_N, {'r': ('T1', r)})
tf_1.set_hps({'alpha': 1.0})
tf_2 = model.DomainInterface(T2_N, {'r': ('T2', r)})
tf_2.set_hps({'alpha': 1.0})
### All one group for everyone
t1_g1 = tf_1.create_group(rng)
for i in range(T1_N):
tf_1.add_entity_to_group(t1_g1, i)
t2_g1 = tf_2.create_group(rng)
for i in range(T2_N):
tf_2.add_entity_to_group(t2_g1, i)
ITERS = 10
for i in range(ITERS):
gibbs.gibbs_sample_type(tf_1, rng)
assert_equal(len(np.unique(tf_1.get_assignments())), 10)
