def test_grad_data3_3(self):
y = self.data3[:, -1:]
X = self.data3[:, :-1]
m, n = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = array([[-24], [0.2], [0.2]])
assert_allclose([[0.043], [2.566], [2.647]],
grad(X, y, theta),
rtol=0,
atol=0.001,
equal_nan=False)
def test_grad_data3_2(self):
y = self.data3[:, -1:]
X = self.data3[:, :-1]
m, _ = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = array([[-0.1], [0.4], [-0.4]])
assert_allclose([[-0.134], [-8.275], [-18.564]],
grad(X, y, theta),
rtol=0,
atol=0.001,
equal_nan=False)
def test_grad_data3_1(self):
y = self.data3[:, -1:]
X = self.data3[:, :-1]
m, n = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = zeros((n + 1, 1), dtype=float64)
assert_allclose([[-0.1000], [-12.0092], [-11.2628]],
grad(X, y, theta),
rtol=0,
atol=5e-05,
equal_nan=False)
def test_grad_data4_1(self):
y = self.data4[:, -1:]
X = self.data4[:, :-1]
m, n = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = zeros((n + 1, 1), dtype=float64)
assert_allclose(array([[0.151], [0.225], [11.154], [9.838], [2.534],
[4.887], [3.733], [0.044], [3.686]]),
grad(X, y, theta),
rtol=0,
atol=0.001,
equal_nan=False)
def test_grad_data4_3(self):
y = self.data4[:, -1:]
X = self.data4[:, :-1]
m, n = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = -0.1 * ones((n + 1, 1), dtype=float64)
assert_allclose(array([[-0.349], [-1.698], [-49.293], [-24.715],
[-7.734], [-35.013], [-12.263], [-0.192],
[-12.935]]),
grad(X, y, theta),
rtol=0,
atol=0.001,
equal_nan=False)
def test_grad_data4_6(self):
y = self.data4[:, -1:]
X = self.data4[:, :-1]
m, n = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = -0.145 * ones((n + 1, 1), dtype=float64)
def J(theta):
return cost_func(X, y, theta)
assert_allclose(grad(X, y, theta),
numerical_grad(J, theta, self.err),
rtol=0,
atol=0.001,
equal_nan=False)
def test_grad_data4_5(self):
y = self.data4[:, -1:]
X = self.data4[:, :-1]
m, _ = X.shape
intercept = ones((m, 1), dtype=float64)
X = append(intercept, X, axis=1)
theta = array([[-0.198], [0.25], [-0.234], [-0.793], [0.123], [-0.378],
[0.423], [-0.678], [-0.3]])
def J(theta):
return cost_func(X, y, theta)
assert_allclose(grad(X, y, theta),
numerical_grad(J, theta, self.err),
rtol=0,
atol=0.001,
equal_nan=False)