def test_steepestdescent_lowiterations():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.steepestdescent(X, y, gradient(),
iterations=2)
target = [0, 0.4]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_steepestdescent_alpha():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.steepestdescent(X, y, gradient(),
alpha=0.001)
target = [0.47, 0.84]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_gradientdescent_stochastic():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.gradientdescent(X, y, gradient(),
stochastic=True)
target = [0.47, 0.84]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_gradientdescent_initialweights():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.gradientdescent(X, y, gradient(),
initial_weights=np.array([0.2, 0.2]))
target = [0.47, 0.84]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_newtonsmethod_alpha():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.newtonsmethod(X, y, gradient(),
hessian(), alpha=0.001)
target = [0.3, 0.5]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_gradientdescent_initialweights():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.gradientdescent(X,
y,
gradient(),
initial_weights=np.array(
[0.2, 0.2]))
target = [0.47, 0.84]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_newtonsmethod_lowiterations():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.newtonsmethod(X,
y,
gradient(),
hessian(),
iterations=2)
target = [0, 0]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_newtonsmethod_alpha():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.newtonsmethod(X,
y,
gradient(),
hessian(),
alpha=0.001)
target = [0.3, 0.5]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_kerneldensityestimate():
x, y = data.continuous_data_complicated()
km = kernelmethods.KernelMethods()
km.fit(x, y)
kde = km.kerneldensityestimate([5], 1)
np.testing.assert_almost_equal(kde, 0)
def test_locallinearregression():
x, y = data.continuous_data_complicated()
km = kernelmethods.KernelMethods()
km.fit(x, y)
predict = km.locallinearregression(5.5, km.gaussiankernel, 0.1)
np.testing.assert_almost_equal(predict, 5.092, 2)
def test_nadarayaaverage():
x, y = data.continuous_data_complicated()
km = kernelmethods.KernelMethods()
km.fit(x, y)
predict = km.nadarayaaverage(5.5, km.gaussiankernel, 0.01)
np.testing.assert_almost_equal(predict, 4.237, 2)
def test_kernelmethods_fit():
x, y = data.continuous_data_complicated()
km = kernelmethods.KernelMethods()
km.fit(x, y)
assert km.learned
def test_steepestdescent_alpha():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.steepestdescent(X, y, gradient(), alpha=0.001)
target = [0.47, 0.84]
np.testing.assert_array_almost_equal(weights, target, 1)
def test_gradientdescent_stochastic():
x, y = data.continuous_data_complicated()
X = np.column_stack((np.ones(np.shape(x)[0]), x))
weights = descentmethods.gradientdescent(X, y, gradient(), stochastic=True)
target = [0.47, 0.84]
np.testing.assert_array_almost_equal(weights, target, 1)
