def test_fit_simple_logit(self):
with self.test_session():
data = PreferenceDict()
a = (0,)
b = (1,)
c = (2,)
d = (3,)
data[a, b] = 1
data[a, c] = 1
data[c, b] = -1
data[c, d] = 1
m = BinaryPreferenceModel(
n_iter=500,
n_samples=20,
link='logit'
)
X, y = preprocess_data(data)
m.fit(X, y)
mean = m.mean([
(0,),
(1,),
(2,),
(3,),
])
self.assertTrue(
all(x > y for x, y in zip(mean, mean[1:]))
)
def test_preprocess_data(self):
pd = UniformPreferenceDict(2)
a = (0, 1)
b = (2, 3)
c = (4, 5)
d = (6, 7)
e = (8, 9)
pd[a, b] = 1
pd[a, c] = 1
pd[a, d] = 1
pd[c, b] = -1
pd[d, e] = 0
X, y = preprocess_data(pd)
A = np.array([
[0, 1],
[2, 3],
[4, 5],
[6, 7],
[8, 9],
])
self.assertTrue(np.allclose(A, X))
b = {
(0, 1): 1,
(0, 2): 1,
(0, 3): 1,
(1, 2): 1,
(3, 4): 0,
}
self.assertEqual(y, b)