def test_order_irrelevance_random_ts(self):
# Build two random series
rs1 = ts.auto_regressive(start_date="2000-01-01",
end_date="2020-01-01",
frequency='M',
start_values=[1.],
cst=1.,
order=1,
coeffs=[0.3],
sigma=0.1)
rs2 = ts.auto_regressive(start_date="2000-01-01",
end_date="2020-01-01",
frequency='M',
start_values=[1.],
cst=1.,
order=1,
coeffs=[0.35],
sigma=0.1)
# Test that order of time series is irrelevant
self.assertEqual(cl.euclidean_distance(rs1, rs2),
cl.euclidean_distance(rs1, rs2))
self.assertEqual(cl.dtw_distance(rs1, rs2), cl.dtw_distance(rs1, rs2))
def test_order_irrelevance_contant_ts(self):
# Build two constant time series
rs1 = ts.constant(start_date="2000-01-01",
end_date="2020-01-01",
frequency='Y',
cst=1.)
rs2 = ts.constant(start_date="2000-01-01",
end_date="2020-01-01",
frequency='Y',
cst=2.)
# Test that order of time series is irrelevant
assert cl.euclidean_distance(rs1,
rs2) == cl.euclidean_distance(rs1, rs2)
assert cl.dtw_distance(rs1, rs2) == cl.dtw_distance(rs1, rs2)
def test_distances_value(self):
# Build two constant time series
rs1 = ts.constant(start_date="2000-01-01",
end_date="2020-01-01",
frequency='Y',
cst=1.)
rs2 = ts.constant(start_date="2000-01-01",
end_date="2020-01-01",
frequency='Y',
cst=2.)
# Test distances are the same
assert cl.euclidean_distance(rs1, rs2) == np.sqrt(rs1.nvalues)
assert cl.dtw_distance(rs1, rs2, mode='abs') == rs1.nvalues
assert cl.dtw_distance(rs1, rs2, mode='square') == np.sqrt(rs1.nvalues)
def test_distances_value(self):
# Build two constant time series
rs1 = ts.constant(start_date="2000-01-01",
end_date="2020-01-01",
frequency='Y',
cst=1.)
rs2 = ts.constant(start_date="2000-01-01",
end_date="2020-01-01",
frequency='Y',
cst=2.)
# Test distances are the same
self.assertEqual(cl.euclidean_distance(rs1, rs2), np.sqrt(rs1.nvalues))
self.assertEqual(cl.dtw_distance(rs1, rs2), np.sqrt(rs1.nvalues))