def test_no_bounds(self):
clf = KMeans()
X = np.zeros(10).reshape(-1, 1)
with self.assertWarns(PrivacyLeakWarning):
clf.fit(X)
def test_1d_array(self):
clf = KMeans()
X = np.zeros(10)
with self.assertRaises(ValueError):
clf.fit(X)
def test_sample_weights(self):
clf = KMeans(3, epsilon=30, bounds=(0, 1))
X = np.array([0.1, 0.1, 0.1, 0.1, 0.5, 0.5, 0.5, 0.5, 0.9, 0.9,
0.9]).reshape(-1, 1)
with self.assertWarns(DiffprivlibCompatibilityWarning):
clf.fit(X, None, 1)
def test_inf_epsilon(self):
global_seed(3141592653)
clf = KMeans(float("inf"), (0, 1), 3)
X = np.array([0.1, 0.1, 0.1, 0.1, 0.5, 0.5, 0.5, 0.5, 0.9, 0.9, 0.9]).reshape(-1, 1)
clf.fit(X)
centers = clf.cluster_centers_
self.assertIn(0.1, centers)
self.assertIn(0.5, centers)
self.assertIn(0.9, centers)
def test_many_features(self):
X = np.random.random(size=(500, 3))
bounds = (0, 1)
clf = KMeans(bounds=bounds, n_clusters=4)
clf.fit(X)
centers = clf.cluster_centers_
self.assertEqual(centers.shape[0], 4)
self.assertEqual(centers.shape[1], 3)
self.assertTrue(np.all(clf.transform(X).argmin(axis=1) == clf.predict(X)))
def test_predict(self):
clf = KMeans(epsilon=1, bounds=(0, 1), n_clusters=3)
X = np.array([0.1, 0.1, 0.1, 0.1, 0.5, 0.5, 0.5, 0.5, 0.9, 0.9, 0.9]).reshape(-1, 1)
clf.fit(X)
predicted = clf.predict(np.array([0.1, 0.5, 0.9]).reshape(-1, 1))
# print("Predictions: %s" % str(predicted))
# print("Centers: %s" % str(clf.cluster_centers_))
self.assertTrue(0 <= predicted[0] <= 2)
self.assertTrue(0 <= predicted[1] <= 2)
self.assertTrue(0 <= predicted[2] <= 2)
def test_predict(self):
clf = KMeans(30, (0, 1), 3)
X = np.array([0.1, 0.1, 0.1, 0.1, 0.5, 0.5, 0.5, 0.5, 0.9, 0.9,
0.9]).reshape(-1, 1)
clf.fit(X)
predicted = clf.predict(np.array([0.1, 0.5, 0.9]).reshape(-1, 1))
# print("Predictions: %s" % str(predicted))
# print("Centers: %s" % str(clf.cluster_centers_))
self.assertNotEqual(predicted[0], predicted[1])
self.assertNotEqual(predicted[0], predicted[2])
self.assertNotEqual(predicted[2], predicted[1])
def test_simple(self):
global_seed(3141592653)
clf = KMeans(epsilon=10, bounds=(0, 1), n_clusters=3)
X = np.zeros(6000) + 0.1
X[:4000] = 0.5
X[:2000] = 0.9
X = X.reshape(-1, 1)
clf.fit(X)
centers = clf.cluster_centers_
self.assertAlmostEqual(np.min(centers), 0.1, delta=0.1)
self.assertAlmostEqual(np.median(centers), 0.5, delta=0.1)
self.assertAlmostEqual(np.max(centers), 0.9, delta=0.1)
def test_simple(self):
global_seed(3141592653)
clf = KMeans(5, (0, 1), 3)
X = np.zeros(1000) + 0.1
X[:666] = 0.5
X[:333] = 0.9
X = X.reshape(-1, 1)
clf.fit(X)
centers = clf.cluster_centers_
self.assertTrue(np.isclose(centers, 0.1, atol=0.05).any())
self.assertTrue(np.isclose(centers, 0.5, atol=0.05).any())
self.assertTrue(np.isclose(centers, 0.9, atol=0.05).any())
def test_accountant(self):
from diffprivlib.accountant import BudgetAccountant
acc = BudgetAccountant()
clf = KMeans(30, (0, 1), 3, accountant=acc)
X = np.array([0.1, 0.1, 0.1, 0.1, 0.5, 0.5, 0.5, 0.5, 0.9, 0.9, 0.9]).reshape(-1, 1)
clf.fit(X)
self.assertEqual((30, 0), acc.total())
clf.fit(X)
self.assertEqual((60, 0), acc.total())
with BudgetAccountant(15, 0) as acc2:
clf2 = KMeans(10, (0, 1), 3)
clf2.fit(X)
self.assertEqual((10, 0), acc2.total())
with self.assertRaises(BudgetError):
clf2.fit(X)