def setUp_dimensions(self):
n = random.randint(100, 1000)
k = random.randint(5, 20)
X = np.random.rand(n, k)
Y = np.random.rand(n, 1)
coef_dim = regression(Y, X)[0].size
X_dim = np.size(X, 1)
se_dim = regression(Y, X)[1].size
return (coef_dim, se_dim, X_dim)
def test_mismatched_obs(self):
# Check error thrown if number of observations different for Y and X
self.setUp()
Y = np.array([[1, 2, 3, 4]]).T
X = np.array([[2, 3, 4], [3, 8, 3]]).T
self.assertFalse(np.size(X, 0) == np.size(Y, 0))
with self.assertRaises(Exception):
coef, se = regression(X, Y)
def test_too_many_vars(self):
# Check error thrown if number of independent variables greater
# than number of observations
self.setUp()
Y = np.array([[1, 2]]).T
X = np.array([[2, 3], [3, 8], [1, 3]]).T
self.assertTrue(np.size(X, 1) > np.size(X, 0))
with self.assertRaises(Exception):
coef, se = regression(X, Y)
def test_novariation(self):
n = random.randint(100, 1000)
k = random.randint(5, 20)
X = np.ones((n, k))
Y = np.ones((n, 1))
try:
coef = regression(Y, X)[0]
error = 0
except np.linalg.LinAlgError:
error = 1
self.assertEqual(error, 1)
