def test_implicit_constant(self):
x = self.rng.standard_normal((1000, 2))
assert not implicit_constant(x)
x[:, 0] = 1.0
assert implicit_constant(x)
x = self.rng.standard_normal((1000, 3))
x[:, 0] = x[:, 0] > 0
x[:, 1] = 1 - x[:, 0]
assert implicit_constant(x)
def test_implicit_constant(self):
x = np.random.standard_normal((1000, 2))
assert_true(not implicit_constant(x))
x[:, 0] = 1.0
assert_true(implicit_constant(x))
x = np.random.standard_normal((1000, 3))
x[:, 0] = x[:, 0] > 0
x[:, 1] = 1 - x[:, 0]
assert_true(implicit_constant(x))
def test_implicit_constant(self):
x = self.rng.standard_normal((1000, 2))
assert not implicit_constant(x)
x[:, 0] = 1.0
assert implicit_constant(x)
x = self.rng.standard_normal((1000, 3))
x[:, 0] = x[:, 0] > 0
x[:, 1] = 1 - x[:, 0]
assert implicit_constant(x)
def _r2(self, params: NDArray) -> float:
y = self._fit_y
x = self._fit_regressors
constant = False
if x is not None and x.shape[1] > 0:
constant = self.constant or implicit_constant(x)
e = self.resids(params)
if constant:
y = y - np.mean(y)
return 1.0 - e.T.dot(e) / y.dot(y)
def _r2(self, params):
y = self._fit_y
x = self._fit_regressors
constant = False
if x is not None and x.shape[1] > 0:
constant = self.constant or implicit_constant(x)
e = self.resids(params)
if constant:
y = y - np.mean(y)
return 1.0 - e.T.dot(e) / y.dot(y)
