def test_linear_baseline_with_spikes_with_first_zeros(self):
"""
Tests that you can fit a linear baseline, when the second 1000 points are
0 and the first 1000 points have a slope.
"""
N = 1010
x = np.arange(0, N, 1)
slope = 2.
offset = 10.
g1 = norm(loc=100, scale=1.0) # generate three gaussian as a signal
g2 = norm(loc=300, scale=3.0)
g3 = norm(loc=750, scale=5.0)
y = np.zeros(N)
y[:1000] = offset + slope * x[:1000] + 100. * g1.pdf(
x[:1000]) + 300. * g2.pdf(x[:1000]) + 500. * g3.pdf(x[:1000])
y[:1000] += np.random.random(1000) * 0.5 - 0.25
z = arPLS(y, lambda_=(10.)**2)
d = y - z
fit = np.polyfit(x[:500], z[:500], 1)
self.assertAlmostEqual(fit[0], slope, places=1) # x
self.assertAlmostEqual(fit[1], offset, places=1) # const
def test_zero_baseline_no_spikes_no_noise(self):
offset = 0.
y = np.zeros(1000) + offset
z = arPLS(y)
d = y - z
self.assertAlmostEqual(np.mean(d), 0, places=1)
def test_same_return_shape(self):
"""
"""
y = np.ones(100)
y += np.random.random(100)
z = arPLS(y)
self.assertEqual(y.shape, z.shape)
def test_zero_baseline_no_spikes(self):
offset = 0.
y = np.zeros(1000) + offset
y += np.random.random(1000) * 0.5 - 0.25
z = arPLS(y)
d = y - z
self.assertAlmostEqual(np.mean(d), 0, places=1)
# test that it has been smoothed
self.assertLess(np.std(z, ddof=1), np.std(y, ddof=1))
def test_linear_baseline_with_spikes(self):
x = np.arange(0, 1000, 1)
slope = 2.
offset = 10.
g1 = norm(loc=100, scale=1.0) # generate three gaussian as a signal
g2 = norm(loc=300, scale=3.0)
g3 = norm(loc=750, scale=5.0)
y = offset + slope * x + 200. * g1.pdf(x) + 300. * g2.pdf(
x) + 500. * g3.pdf(x)
y += np.random.random(1000) * 0.5 - 0.25
z = arPLS(y)
d = y - z
fit = np.polyfit(x, z, 2)
self.assertAlmostEqual(fit[0], 0, places=1) # x**2
self.assertAlmostEqual(fit[1], slope, places=1) # x
self.assertAlmostEqual(fit[2], offset, places=1) # const
def test_linear_baseline_no_spikes(self):
x = np.arange(0, 1000, 1)
slope = 2.
offset = 10.
y = offset + slope * x
y += np.random.random(1000) * 0.5 - 0.25
z = arPLS(y)
d = y - z
fit = np.polyfit(x, z, 2)
self.assertAlmostEqual(fit[0], 0, places=1) # x**2
self.assertAlmostEqual(fit[1], slope, places=1) # x
self.assertAlmostEqual(fit[2] - offset, 0., places=1) # const
f = np.poly1d(fit)
y_sub = y - f(x)
z_sub = z - f(x)
# test that it is smoother
self.assertAlmostEqual(np.mean(z_sub), 0., places=3)
self.assertLess(np.std(z_sub, ddof=1), np.std(y_sub, ddof=1))
def remove_baseline(self, ydata):
baseline = spyctra.arPLS(ydata, lambda_=1.e3)
diff = np.array(ydata) - np.array(baseline)
return baseline, diff