def test_nan_and_none():
# deal with NaN, None?
c, m = _transition_counts([[0, np.nan]])
assert m == {0: 0}
np.testing.assert_array_equal(c, np.zeros((1, 1)))
c, m = _transition_counts([[np.nan]])
assert m == {}
np.testing.assert_array_equal(c, np.zeros((0, 0)))
if not PY3:
c, m = _transition_counts([[None, None]])
assert m == {}
np.testing.assert_array_equal(c, np.zeros((0, 0)))
def test_nan_and_none():
# deal with NaN, None?
c, m = _transition_counts([[0, np.nan]])
assert m == {0: 0}
np.testing.assert_array_equal(c, np.zeros((1, 1)))
c, m = _transition_counts([[np.nan]])
assert m == {}
np.testing.assert_array_equal(c, np.zeros((0, 0)))
if not PY3:
c, m = _transition_counts([[None, None]])
assert m == {}
np.testing.assert_array_equal(c, np.zeros((0, 0)))
def test_big_counts():
# try using really big numbers, and we still want a small transition matrix
c, m = _transition_counts([[100000000, 100000000, 100000001, 100000001]])
np.testing.assert_array_equal(c, 1.0 * np.array([
[1, 1],
[0, 1],
]))
assert m == {100000000: 0, 100000001: 1}
def test_big_counts():
# try using really big numbers, and we still want a small transition matrix
c, m = _transition_counts([[100000000, 100000000, 100000001, 100000001]])
np.testing.assert_array_equal(c, 1.0 * np.array([
[1, 1],
[0, 1],
]))
assert m == {100000000: 0, 100000001: 1}
def test_string_labels():
# try using strings as labels
c, m = _transition_counts([['alpha', 'b', 'b', 'b', 'c']])
np.testing.assert_array_equal(c, 1.0 * np.array([
[0, 1, 0],
[0, 2, 1],
[0, 0, 0]
]))
assert m == {'alpha': 0, 'b': 1, 'c': 2}
def test_argument():
# test that first argument must be a list of sequences
with np.testing.assert_raises(ValueError):
_transition_counts([1, 2, 3])
def test_no_counts():
c, m = _transition_counts([[0]])
def test_string_labels():
# try using strings as labels
c, m = _transition_counts([['alpha', 'b', 'b', 'b', 'c']])
np.testing.assert_array_equal(
c, 1.0 * np.array([[0, 1, 0], [0, 2, 1], [0, 0, 0]]))
assert m == {'alpha': 0, 'b': 1, 'c': 2}
def test_lag_time():
# test the simple example with lag_time > 1
c, m = _transition_counts([range(10)], lag_time=2)
np.testing.assert_array_equal(c, 0.5 * np.eye(10, k=2))
def test_lag_time():
# test the simple example with lag_time > 1
c, m = _transition_counts([range(10)], lag_time=2)
np.testing.assert_array_equal(c, 0.5 * np.eye(10, k=2))
def test_sliding_window():
X = np.arange(10)
C1, m1 = _transition_counts([X], lag_time=3, sliding_window=False)
C2, m2 = _transition_counts([X[::3]], sliding_window=True)
np.testing.assert_array_almost_equal(C1, C2)
assert m1 == m2
def test_lag_time_norm():
X = np.arange(6)
C, _ = _transition_counts([X], lag_time=3)
np.testing.assert_array_almost_equal(C, np.eye(6, k=3) / 3)
def test_argument():
# test that first argument must be a list of sequences
with np.testing.assert_raises(ValueError):
_transition_counts([1, 2, 3])
def test_no_counts():
c, m = _transition_counts([[0]])
def test_lag_time_norm():
X = np.arange(6)
C, _ = _transition_counts([X], lag_time=3)
np.testing.assert_array_almost_equal(C, np.eye(6, k=3) / 3)
def test_upper_triangular():
# test a simple example
c, m = _transition_counts([np.arange(10)])
np.testing.assert_array_equal(c, np.eye(10, k=1))
assert list(m.keys()) == list(range(10))
assert list(m.values()) == list(range(10))
def test_sliding_window():
X = np.arange(10)
C1, m1 = _transition_counts([X], lag_time=3, sliding_window=False)
C2, m2 = _transition_counts([X[::3]], sliding_window=True)
np.testing.assert_array_almost_equal(C1, C2)
assert m1 == m2
def test_upper_triangular():
# test a simple example
c, m = _transition_counts([np.arange(10)])
np.testing.assert_array_equal(c, np.eye(10, k=1))
assert list(m.keys()) == list(range(10))
assert list(m.values()) == list(range(10))
