def test_get_critical_set_indices_single(self):
x = np.concatenate([self.x, np.array([self.x[0, :]] * 5)], axis=0)
x_min = x[:1, :]
x_maj = x[1:, :]
critical_set = BiasedRFClassifier(k_nearest_neighbor=5)._get_crtical_set_indices(x_maj, x_min)
# confirm that the indices are the nearest neighbors
self.assertEqual(list(range(9, 14)), critical_set)
def test_get_critical_x_shape(self):
x = np.concatenate([self.x, np.array([self.x[0, :]] * 5)], axis=0)
y = np.zeros((len(x), 1))
y[0] = 1
_, critical_y = BiasedRFClassifier(k_nearest_neighbor=5)._get_critical_set(x, y)
# confirm the critical y is the minor (label=1) and 5 major (label=0) stacked
self.assertListEqual([1,0,0,0,0,0], critical_y.tolist())
def test_get_critical_x_shape(self):
x = np.concatenate([self.x, np.array([self.x[0, :]] * 5)], axis=0)
y = np.zeros((len(x), 1))
y[0] = 1
critical_x, _ = BiasedRFClassifier(k_nearest_neighbor=5)._get_critical_set(x, y)
# confirm getting the single minor and 5 cloeset => total 6 rows back
self.assertEqual(6, critical_x.shape[0])
self.assertEqual(x.shape[1], critical_x.shape[1])
def test_get_critical_set_indices_in_major_set(self):
x_min = self.x[:2, :]
x_maj = self.x[2:, :]
critical_set = BiasedRFClassifier(k_nearest_neighbor=5)._get_crtical_set_indices(x_maj, x_min)
# confirm that the indices are the nearest neighbors
major_indices = list(range(len(x_maj)))
for idx in critical_set:
with self.subTest(idx=idx):
self.assertTrue(idx in major_indices)
def test_get_critical_set_indices_overlap_neighbors(self):
# create dataset where x_min are two points with the same features
# so the critical set we get should just be 5 points (both have the same neighbors)
self.x[1, :] = self.x[0, :]
x_min = self.x[:2, :]
x_maj = self.x[2:, :]
critical_set = BiasedRFClassifier(k_nearest_neighbor=5)._get_crtical_set_indices(x_maj, x_min)
# confirm that the indices are the nearest neighbors
self.assertEqual(5, len(critical_set))
def test_predict_logic(self):
model = BiasedRFClassifier(k_nearest_neighbor=5)
model.fit(self.x, self.imbalance_y)
pred = model.predict(self.x)
self.assertEqual(len(self.imbalance_y), len(pred))
print(pred)
def test_fit_logic(self):
model = BiasedRFClassifier(k_nearest_neighbor=5)
model.fit(self.x, self.imbalance_y)
self.assertTrue(model._is_fitted())