def test_apply_kernel_linear(self):
"""Check linear kernel is applied correctly."""
clf = BaseRVM(kernel='linear', bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[17, 23], [39, 53]])
np.testing.assert_array_equal(phi, target)
def test_apply_kernel_bias(self):
"""Check the kernel function correctly applies a bias."""
clf = BaseRVM(kernel='linear', bias_used=True)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[17, 23, 1], [39, 53, 1]])
np.testing.assert_array_equal(phi, target)
def test_apply_kernel_rbf(self):
"""Check RBF kernel is applied correctly."""
clf = BaseRVM(kernel='rbf', bias_used=False, coef1=0.5)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[1.12535175e-07, 2.31952283e-16],
[1.83156389e-02, 1.12535175e-07]])
np.testing.assert_allclose(phi, target)
def test_apply_kernel_invalid(self):
"""Check that an invalid kernel choice raises an exception."""
clf = BaseRVM(kernel='wrong')
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
try:
clf._apply_kernel(x, y)
except ValueError as error:
self.assertEqual(str(error), "Kernel selection is invalid.")
else:
self.fail()
def setUp(self):
"""Set up some common test instances."""
self.param_test_clf = BaseRVM(kernel='linear',
degree=1,
coef1=2,
coef0=3,
n_iter=200,
tol=1e-1,
alpha=1e-2,
threshold_alpha=1e3,
beta=1e-3,
beta_fixed=True,
bias_used=False,
verbose=True)
def test_apply_kernel_custom(self):
"""Check custom kernels are applied correctly."""
def custom(x, y):
return 2 * x.dot(y.T)
clf = BaseRVM(kernel=custom, bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[34, 46], [78, 106]])
np.testing.assert_array_equal(phi, target)
def test_apply_kernel_poly(self):
"""Check polynomial kernel is applied correctly."""
clf = BaseRVM(kernel='poly',
bias_used=False,
degree=2,
coef1=1,
coef0=0.5)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[306.25, 552.25], [1560.25, 2862.25]])
np.testing.assert_allclose(phi, target)
def test_apply_kernel_1D(self):
"""Check that _apply_kernel catches a non-2D phi."""
def custom(x, y):
return np.array([1, 2])
clf = BaseRVM(kernel=custom, bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
try:
clf._apply_kernel(x, y)
except ValueError as error:
self.assertEqual(
str(error), "Custom kernel function did not return 2D matrix")
else:
self.fail()
def test_apply_kernel_row_mismatch(self):
"""Check that _apply_kernel catches a mismatch between input/output."""
def custom(x, y):
return np.array([[1, 2]])
clf = BaseRVM(kernel=custom, bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
try:
clf._apply_kernel(x, y)
except ValueError as error:
self.assertEqual(
str(error),
"""Custom kernel function did not return matrix with rows"""
""" equal to number of data points.""")
else:
self.fail()
class BaseRVMTestCase(TestCase):
"""Automatic tests for the BaseRVM class."""
def setUp(self):
"""Set up some common test instances."""
self.param_test_clf = BaseRVM(kernel='linear',
degree=1,
coef1=2,
coef0=3,
n_iter=200,
tol=1e-1,
alpha=1e-2,
threshold_alpha=1e3,
beta=1e-3,
beta_fixed=True,
bias_used=False,
verbose=True)
def test__init__(self):
"""Check parameters are initialized correctly."""
self.assertEqual(self.param_test_clf.kernel, 'linear')
self.assertEqual(self.param_test_clf.degree, 1)
self.assertEqual(self.param_test_clf.coef1, 2)
self.assertEqual(self.param_test_clf.coef0, 3)
self.assertEqual(self.param_test_clf.n_iter, 200)
self.assertEqual(self.param_test_clf.tol, 1e-1)
self.assertEqual(self.param_test_clf.alpha, 1e-2)
self.assertEqual(self.param_test_clf.threshold_alpha, 1e3)
self.assertEqual(self.param_test_clf.beta, 1e-3)
self.assertTrue(self.param_test_clf.beta_fixed)
self.assertFalse(self.param_test_clf.bias_used)
self.assertTrue(self.param_test_clf.verbose)
def test_get_params(self):
"""Check get_params returns a dictionary of the params."""
params = self.param_test_clf.get_params()
self.assertEqual(
params, {
'kernel': 'linear',
'degree': 1,
'coef1': 2,
'coef0': 3,
'n_iter': 200,
'tol': 1e-1,
'alpha': 1e-2,
'threshold_alpha': 1e3,
'beta': 1e-3,
'beta_fixed': True,
'bias_used': False,
'verbose': True
})
def test_set_params(self):
"""Check that set_params sets params from dictionary."""
params = {
'kernel': 'poly',
'degree': 4,
'coef1': 5,
'coef0': 6,
'n_iter': 100,
'tol': 1e-4,
'alpha': 1e-5,
'threshold_alpha': 1e6,
'beta': 1e-6,
'beta_fixed': False,
'bias_used': True,
'verbose': False
}
self.param_test_clf.set_params(**params)
self.assertEqual(self.param_test_clf.kernel, 'poly')
self.assertEqual(self.param_test_clf.degree, 4)
self.assertEqual(self.param_test_clf.coef1, 5)
self.assertEqual(self.param_test_clf.coef0, 6)
self.assertEqual(self.param_test_clf.n_iter, 100)
self.assertEqual(self.param_test_clf.tol, 1e-4)
self.assertEqual(self.param_test_clf.alpha, 1e-5)
self.assertEqual(self.param_test_clf.threshold_alpha, 1e6)
self.assertEqual(self.param_test_clf.beta, 1e-6)
self.assertFalse(self.param_test_clf.beta_fixed)
self.assertTrue(self.param_test_clf.bias_used)
self.assertFalse(self.param_test_clf.verbose)
def test_apply_kernel_linear(self):
"""Check linear kernel is applied correctly."""
clf = BaseRVM(kernel='linear', bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[17, 23], [39, 53]])
np.testing.assert_array_equal(phi, target)
def test_apply_kernel_rbf(self):
"""Check RBF kernel is applied correctly."""
clf = BaseRVM(kernel='rbf', bias_used=False, coef1=0.5)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[1.12535175e-07, 2.31952283e-16],
[1.83156389e-02, 1.12535175e-07]])
np.testing.assert_allclose(phi, target)
def test_apply_kernel_poly(self):
"""Check polynomial kernel is applied correctly."""
clf = BaseRVM(kernel='poly',
bias_used=False,
degree=2,
coef1=1,
coef0=0.5)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[306.25, 552.25], [1560.25, 2862.25]])
np.testing.assert_allclose(phi, target)
def test_apply_kernel_custom(self):
"""Check custom kernels are applied correctly."""
def custom(x, y):
return 2 * x.dot(y.T)
clf = BaseRVM(kernel=custom, bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[34, 46], [78, 106]])
np.testing.assert_array_equal(phi, target)
def test_apply_kernel_bias(self):
"""Check the kernel function correctly applies a bias."""
clf = BaseRVM(kernel='linear', bias_used=True)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
phi = clf._apply_kernel(x, y)
target = np.array([[17, 23, 1], [39, 53, 1]])
np.testing.assert_array_equal(phi, target)
def test_apply_kernel_invalid(self):
"""Check that an invalid kernel choice raises an exception."""
clf = BaseRVM(kernel='wrong')
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
try:
clf._apply_kernel(x, y)
except ValueError as error:
self.assertEqual(str(error), "Kernel selection is invalid.")
else:
self.fail()
def test_apply_kernel_1D(self):
"""Check that _apply_kernel catches a non-2D phi."""
def custom(x, y):
return np.array([1, 2])
clf = BaseRVM(kernel=custom, bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
try:
clf._apply_kernel(x, y)
except ValueError as error:
self.assertEqual(
str(error), "Custom kernel function did not return 2D matrix")
else:
self.fail()
def test_apply_kernel_row_mismatch(self):
"""Check that _apply_kernel catches a mismatch between input/output."""
def custom(x, y):
return np.array([[1, 2]])
clf = BaseRVM(kernel=custom, bias_used=False)
x = np.array([[1, 2], [3, 4]])
y = np.array([[5, 6], [7, 8]])
try:
clf._apply_kernel(x, y)
except ValueError as error:
self.assertEqual(
str(error),
"""Custom kernel function did not return matrix with rows"""
""" equal to number of data points.""")
else:
self.fail()
