def test_MBGD6(self, err):
def grad(X, y, theta, schleem, plumbus, wubba, lubba):
m = len(y)
grad = (schleem / (m * wubba))
grad = grad * (X.T).dot(X.dot(theta) - y)
grad = grad + plumbus / (2 * lubba)
return grad
X = array([[0, 1, 2], [-1, 5, 3.6], [2, 0.5, 1],
[2, -1, 4], [3, 6, -3], [0, 7.8, 3.5],
[2.5, 3, 4.3], [3.2, 5.7, -3], [0, 7.8, 3.5], [9, 8, 7]])
initial_theta = array([[3], [-2], [0.7]])
y = array([[0.3], [1.2], [0.5], [0.8], [1.5],
[-0.75], [0.43], [0.62], [0.85], [-0.3]])
num_iters = 5
plumbus = 1.2
schleem = 0.9
wubba = 2.4
lubba = 3
alpha = 0.1
b = 5
assert_allclose(array([[-0.0062510], [-0.3414393], [-0.3127686]]),
MBGD(X, y, grad, initial_theta,
alpha, num_iters, b, lubba=lubba,
schleem=schleem, wubba=wubba,
plumbus=plumbus),
rtol=0, atol=0.001, equal_nan=False)
def test_MBGD4(self, err):
def grad(X, y, theta):
m = len(y)
grad = (1 / m) * (X.T).dot(X.dot(theta) - y)
return grad
X = array([[0, 1, 2], [-1, 5, 3.6], [2, 0.5, 1],
[2, -1, 4], [3, 6, -3], [0, 7.8, 3.5],
[2.5, 3, 4.3], [3.2, 5.7, -3], [0, 7.8, 3.5], [9, 8, 7]])
initial_theta = array([[3], [-2], [0.7]])
y = array([[0.3], [1.2], [0.5], [0.8], [1.5],
[-0.75], [0.43], [0.62], [0.85], [-0.3]])
num_iters = 1
alpha = 0.05
b = 5
assert_allclose(array([[0.67576], [-2.01056], [-0.80705]]),
MBGD(X, y, grad, initial_theta,
alpha, num_iters, b),
rtol=0, atol=0.001, equal_nan=False)