def test_transform(self):
a = LocalEncodedBond()
a.fit(ALL_DATA)
m = a.transform(ALL_DATA)
expected_results = numpy.array(
[17.068978019300587, 54.629902544876572, 1006.4744899075993])
mm = numpy.array([x.sum() for x in m])
self.assertTrue((numpy.allclose(mm, expected_results)))
def test_transform(self):
a = LocalEncodedBond()
a.fit(ALL_DATA)
m = a.transform(ALL_DATA)
expected_results = numpy.array([17.068978019300587,
54.629902544876572,
1006.4744899075993])
mm = numpy.array([x.sum() for x in m])
self.assertTrue((numpy.allclose(mm, expected_results)))
def test_transform_max_depth1(self):
a = LocalEncodedBond(max_depth=1)
a.fit(ALL_DATA)
m = a.transform(ALL_DATA)
expected_results = numpy.array([6.82758723, 6.82758018, 88.75860423])
mm = numpy.array([x.sum() for x in m])
try:
numpy.testing.assert_allclose(mm, expected_results)
except AssertionError as e:
self.fail(e)
def test_get_labels(self):
a = LocalEncodedBond(segments=2, start=0., end=1.)
m = a.fit_transform([METHANE])
labels = a.get_labels()
self.assertEqual(m.shape[2], len(labels))
expected = (
'C_0.0', 'C_1.0',
'H_0.0', 'H_1.0',
)
self.assertEqual(labels, expected)
def test_transform_max_depth1(self):
a = LocalEncodedBond(max_depth=1)
a.fit(ALL_DATA)
m = a.transform(ALL_DATA)
expected_results = numpy.array([6.82758723,
6.82758018,
88.75860423])
mm = numpy.array([x.sum() for x in m])
try:
numpy.testing.assert_allclose(mm, expected_results)
except AssertionError as e:
self.fail(e)
def test_get_labels(self):
a = LocalEncodedBond(segments=2, start=0., end=1.)
m = a.fit_transform([METHANE])
labels = a.get_labels()
self.assertEqual(m.shape[2], len(labels))
expected = (
'C_0.0',
'C_1.0',
'H_0.0',
'H_1.0',
)
self.assertEqual(labels, expected)
def test_small_to_large(self):
a = LocalEncodedBond()
a.fit([METHANE])
# This is a cheap test to prevent needing all the values here
expected_results = numpy.array([
0.016125813269, # mean
0.065471987297, # std
0., # min
0.398807098298, # max
29.02646388512, # sum
])
try:
m = a.transform([MID])
val = numpy.array([
m.mean(),
m.std(),
m.min(),
m.max(),
m.sum(),
])
numpy.testing.assert_allclose(val, expected_results)
except AssertionError as e:
self.fail(e)
def test_small_to_large(self):
a = LocalEncodedBond()
a.fit([METHANE])
# This is a cheap test to prevent needing all the values here
expected_results = numpy.array([
0.016125813269, # mean
0.065471987297, # std
0., # min
0.398807098298, # max
29.02646388512, # sum
])
try:
m = a.transform([MID])
val = numpy.array([
m.mean(),
m.std(),
m.min(),
m.max(),
m.sum(),
])
numpy.testing.assert_allclose(val, expected_results)
except AssertionError as e:
self.fail(e)
def test_fit(self):
a = LocalEncodedBond()
a.fit(ALL_DATA)
self.assertEqual(a._elements, set(['N', 'C', 'O', 'H']))
def test_form(self):
a = LocalEncodedBond(form=0)
m = a.fit_transform([METHANE])
self.assertEqual(m.shape, (1, 5, 100))
def test_add_unknown(self):
a = LocalEncodedBond(add_unknown=True)
a.fit([METHANE])
m = a.transform([MID])
self.assertEqual(m.shape, (1, 9, 300))
def test_transform_invalid_spacing(self):
a = LocalEncodedBond(spacing='not real"')
with self.assertRaises(KeyError):
a.fit_transform(ALL_DATA)
def test_transform_before_fit(self):
a = LocalEncodedBond()
with self.assertRaises(ValueError):
a.transform(ALL_DATA)
def test_transform_before_fit(self):
a = LocalEncodedBond()
with self.assertRaises(ValueError):
a.transform(ALL_DATA)
def test_fit(self):
a = LocalEncodedBond()
a.fit(ALL_DATA)
self.assertEqual(a._elements, (('C', ), ('H', ), ('N', ), ('O', )))
def test_fit(self):
a = LocalEncodedBond()
a.fit(ALL_DATA)
self.assertEqual(a._elements, ('C', 'H', 'N', 'O'))
def test_form(self):
a = LocalEncodedBond(form=0)
m = a.fit_transform([METHANE])
self.assertEqual(m.shape, (1, 5, 100))
def test_add_unknown(self):
a = LocalEncodedBond(add_unknown=True)
a.fit([METHANE])
m = a.transform([MID])
self.assertEqual(m.shape, (1, 9, 300))
def test_transform_invalid_spacing(self):
a = LocalEncodedBond(spacing='not real"')
with self.assertRaises(KeyError):
a.fit_transform(ALL_DATA)