Issue413 sampletau (blue-yonder#500)

* refactoring drifbif_simulation.py
* remove test_tsfresh_baseline_dataset
* update threshold in test_ratio to 0.0006, refactoring

tests/integrations/examples/test_driftbif_simulation.py

@@ -14,45 +14,44 @@ def test_intrinsic_velocity_at_default_bifurcation_point(self):
         """
         The intrinsic velocity of a dissipative soliton at the Drift bifurcation point is zero.
         """
-        ds = velocity(tau=1.0/0.3)
+        ds = velocity(tau=1.0 / 0.3)
         self.assertEqual(ds.deterministic, 0.0)
 
     def test_relaxation_dynamics(self):
         """
         Test accuracy of integrating the deterministic dynamics [6, p. 116]
         """
-        ds = velocity(tau=1.01/0.3, R=0)
+        ds = velocity(tau=1.01 / 0.3, R=0)
         v0 = 1.01 * ds.deterministic
 
-        Nt = 100 # Number of time steps
+        Nt = 100  # Number of time steps
         v = ds.simulate(Nt, v0=np.array([v0, 0.]))
 
         k3t = ds.kappa_3 * ds.tau
-        k3st = ds.kappa_3**2 * ds.tau
+        k3st = ds.kappa_3 ** 2 * ds.tau
         a0 = v0 / ds.kappa_3
         acceleration = lambda t: ds.kappa_3 * (a0 * np.sqrt(k3t - 1) * np.exp(k3st * t) /
-                                               np.sqrt(np.exp(2.0 * k3st * t) * ds.Q * a0**2 +
-                                                       np.exp(2.0 * ds.kappa_3 * t) * (k3t - 1 - ds.Q * a0**2)))
+                                               np.sqrt(np.exp(2.0 * k3st * t) * ds.Q * a0 ** 2 +
+                                                       np.exp(2.0 * ds.kappa_3 * t) * (k3t - 1 - ds.Q * a0 ** 2)))
         t = ds.delta_t * np.arange(Nt)
-        return np.testing.assert_array_almost_equal(v[:,0], np.vectorize(acceleration)(t),
+        return np.testing.assert_array_almost_equal(v[:, 0], np.vectorize(acceleration)(t),
                                                     decimal=8)
 
     def test_equlibrium_velocity(self):
         """
         Test accuracy of integrating the deterministic dynamics for equilibrium velocity [6, p. 116]
         """
-        ds = velocity(tau=1.01/0.3, R=0)
+        ds = velocity(tau=1.01 / 0.3, R=0)
         v0 = ds.deterministic
 
-        Nt = 100 # Number of time steps
+        Nt = 100  # Number of time steps
         v = ds.simulate(Nt, v0=np.array([v0, 0.]))
 
-        return np.testing.assert_array_almost_equal(v[:,0]-v0, np.zeros(Nt),
+        return np.testing.assert_array_almost_equal(v[:, 0] - v0, np.zeros(Nt),
                                                     decimal=8)
 
-
     def test_dimensionality(self):
-        ds = velocity(tau=1.0/0.3)
+        ds = velocity(tau=1.0 / 0.3)
         Nt = 10
         v = ds.simulate(Nt)
         self.assertEqual(v.shape, (Nt, 2),
@@ -62,6 +61,7 @@ def test_dimensionality(self):
         self.assertEqual(v.shape, (Nt, 3),
                          'The returned vector should reflect the dimension of the initial condition.')
 
+
 class SampleTauTestCase(unittest.TestCase):
     def test_range(self):
         tau = sample_tau(100)
@@ -71,14 +71,15 @@ def test_range(self):
     def test_ratio(self):
         tau = sample_tau(100000, ratio=0.4)
         sample = np.array(tau)
-        before = np.sum(sample <= 1/0.3)
-        beyond = np.sum(sample > 1/0.3)
-        self.assertTrue(abs(0.4 - float(before)/(before+beyond))<0.002)
-
+        before = np.sum(sample <= 1 / 0.3)
+        beyond = np.sum(sample > 1 / 0.3)
+        self.assertTrue(abs(0.4 - float(before) / (before + beyond)) < 0.006)
+
+
 class LoadDriftBifTestCase(unittest.TestCase):
     def test_classification_labels(self):
         X, y = load_driftbif(10, 100)
-        self.assertEqual(set(y), set([0,1]))
+        self.assertEqual(set(y), set([0, 1]))
 
     def test_regression_labels(self):
         Nsamples = 10
@@ -97,6 +98,3 @@ def test_configured_dimensionality(self):
         Nt = 100
         X, y = load_driftbif(Nsamples, Nt, m=3)
         self.assertEqual(X.shape, (3 * Nt * Nsamples, 4))
-
-if __name__ == '__main__':
-    unittest.main()

tsfresh/examples/driftbif_simulation.py

@@ -62,7 +62,7 @@ def __init__(self, tau=3.8, kappa_3=0.3, Q=1950.0, R=3e-4, delta_t=0.05, seed=No
 
         if not seed is None:
             np.random.seed(seed)
-            
+
         if tau <= 1.0 / kappa_3:
             self.deterministic = 0.0
         else:
@@ -91,9 +91,9 @@ def simulate(self, N, v0=np.zeros(2)):
         :rtype: ndarray
         """
 
-        v = [v0]                        # first value is initial condition
-        n = N - 1                       # Because we are returning the initial condition,
-                                        # only (N-1) time steps are computed
+        v = [v0]  # first value is initial condition
+        n = N - 1  # Because we are returning the initial condition,
+        # only (N-1) time steps are computed
         gamma = np.random.randn(n, v0.size)
         for i in range(n):
             next_v = self.__call__(v[i]) + self.c * gamma[i]
@@ -120,12 +120,13 @@ def sample_tau(n=10, kappa_3=0.3, ratio=0.5, rel_increase=0.15):
     assert ratio > 0 and ratio <= 1
     assert kappa_3 > 0
     assert rel_increase > 0 and rel_increase <= 1
-    tau_c = 1.0/kappa_3
-    
+    tau_c = 1.0 / kappa_3
+
     tau_max = tau_c * (1.0 + rel_increase)
-    tau = tau_c + (tau_max - tau_c) * (np.random.rand(n)-ratio)
+    tau = tau_c + (tau_max - tau_c) * (np.random.rand(n) - ratio)
     return tau.tolist()
 
+
 def load_driftbif(n, l, m=2, classification=True, kappa_3=0.3, seed=False):
     """
     Simulates n time-series with l time steps each for the m-dimensional velocity of a dissipative soliton
@@ -180,5 +181,5 @@ def load_driftbif(n, l, m=2, classification=True, kappa_3=0.3, seed=False):
     df = pd.DataFrame({'id': id, "time": time, "value": np.stack(values).flatten(), "dimension": dimensions})
     y = pd.Series(labels)
     y.index = range(n)
-    
+
     return df, y