def test_interp():
# univariate time series
N = 100
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert np.ndim(Xc[0]) == 1
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert np.ndim(Xc[0]) == 1
assert np.all(np.isin(yc, np.arange(6)))
# multivariate time series
N = 100
D = 5
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N, D)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert Xc[0].shape[1] == D
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert Xc[0].shape[1] == D
assert np.all(np.isin(yc, np.arange(6)))
def test_interp():
N = 100
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert np.all(np.isin(yc, np.arange(6)))
def test_interp():
# univariate time series
N = 100
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert np.ndim(Xc[0]) == 1
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert np.ndim(Xc[0]) == 1
assert np.all(np.isin(yc, np.arange(6)))
# multivariate time series
N = 100
D = 5
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N, D)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert Xc[0].shape[1] == D
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert Xc[0].shape[1] == D
assert np.all(np.isin(yc, np.arange(6)))
# sorting case
N = 100
t = np.arange(N)
t[0:3] = 0
X = [np.column_stack([t, np.random.rand(N)])]
y = [np.random.rand(N)]
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
interp = transform.Interp(sample_period=2, assume_sorted=False)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(w) == 2
assert issubclass(w[-1].category, UserWarning)
assert "duplicate" in str(w[-1].message)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert np.ndim(Xc[0]) == 1
assert np.count_nonzero(np.isnan(Xc)) == 0
def test_interp():
# univariate time series
N = 100
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert np.ndim(Xc[0]) == 1
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert np.ndim(Xc[0]) == 1
assert np.all(np.isin(yc, np.arange(6)))
# multivariate time series
N = 100
D = 5
t = np.arange(N) + np.random.rand(N)
X = [np.column_stack([t, np.random.rand(N, D)])]
y = [np.random.rand(N)]
interp = transform.Interp(2)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert Xc[0].shape[1] == D
y = [np.random.randint(0, 5, N)]
interp = transform.Interp(5, kind='cubic', categorical_target=True)
interp.fit(X, y)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 5
assert len(yc[0]) == N / 5
assert Xc[0].shape[1] == D
assert np.all(np.isin(yc, np.arange(6)))
# sorthing case
N = 100
t = np.arange(N)
t[0:3] = 0
X = [np.column_stack([t, np.random.rand(N)])]
y = [np.random.rand(N)]
interp = transform.Interp(sample_period=2, assume_sorted=False)
interp.fit(X)
Xc, yc, swt = interp.transform(X, y)
assert len(Xc[0]) == N / 2
assert len(yc[0]) == N / 2
assert np.ndim(Xc[0]) == 1
assert np.count_nonzero(np.isnan(Xc)) == 0