def model_builder(self):
"""
Build a KMeans model with the class params.
# Returns:
model: KMeans model.
"""
model = KMeansModel(n_clusters=self._num_clusters,
n_features=self._num_features)
return model
def test_get_model_params(mock_kmeans):
model = Mock()
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
model_params = kmm.get_model_params()
assert np.array_equal(kmm._model.cluster_centers_, model_params)
def test_train(mock_kmeans):
model = Mock()
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
data = np.random.rand(10)
kmm.train(data)
model.fit.assert_called_once_with(data)
def test_predict(mock_kmeans):
model = Mock()
y_pred_real = np.random.rand(10)
model.predict.return_value = y_pred_real
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
data = np.random.rand(10)
y_pred = kmm.predict(data)
model.predict.assert_called_once_with(data)
assert np.array_equal(y_pred_real, y_pred)
def test_kmeans_model_ndarray(mock_kmeans):
model = Mock()
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = np.array([0, 0])
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
mock_kmeans.assert_called_once_with(n_clusters=n_clusters,
init=init,
n_init=n_init)
assert model == kmm._model
assert np.array_equal(init, kmm._init)
assert n_init == kmm._n_init
assert n_features == kmm._n_features
assert np.array_equal(init, kmm._model.cluster_centers_)
def test_set_model_params(mock_kmeans):
model = Mock()
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 1 # Explicit initial center position passed: performing only one init
kmm = KMeansModel(n_clusters, n_features, init, n_init)
params = np.random.rand(10).reshape((5, 2))
kmm.set_model_params(params)
assert model == kmm._model
assert np.array_equal(params, kmm._init)
assert n_init == kmm._n_init
assert n_features == kmm._n_features
assert np.array_equal(params, kmm._model.cluster_centers_)
def test_set_model_params_zeros_array(mock_kmeans):
model = Mock()
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
params = np.zeros(10).reshape((5, 2))
kmm.set_model_params(params)
assert model == kmm._model
assert init == kmm._init
assert n_init == kmm._n_init
assert n_features == kmm._n_features
assert np.array_equal(np.zeros((params.shape[0], n_features)),
kmm._model.cluster_centers_)
def test_performance(mock_kmeans, mock_v_measure_score):
model = Mock()
mock_kmeans.return_value = model
mock_v_measure_score.return_value = 0
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
data = np.random.rand(25).reshape((5, 5))
labels = np.random.randint(0, 2, 5)
kmm.predict = Mock()
prediction = ~labels
kmm.predict.return_value = prediction
res = kmm.performance(data, labels)
assert res == 0
mock_v_measure_score.assert_called_once_with(labels, prediction)
kmm.predict.assert_called_once_with(data)
def test_evaluate(mock_adjusted_rand_score, mock_v_measure_score,
mock_completeness_score, mock_homogeneity_score,
mock_kmeans):
mock_adjusted_rand_score.return_value = 3
mock_v_measure_score.return_value = 2
mock_completeness_score.return_value = 1
mock_homogeneity_score.return_value = 0
model = Mock()
y_pred_real = np.random.rand(10)
model.predict.return_value = y_pred_real
mock_kmeans.return_value = model
n_clusters = 5
n_features = 5
init = 'k-means++'
n_init = 10
kmm = KMeansModel(n_clusters, n_features, init, n_init)
kmm.predict = Mock()
y_pred = np.random.randint(0, 2, 10)
kmm.predict.return_value = y_pred
data = np.random.rand(10)
labels = np.random.randint(0, 2, 10)
h**o, compl, v_meas, rai = kmm.evaluate(data, labels)
kmm.predict.assert_called_once_with(data)
mock_homogeneity_score.assert_called_once_with(labels, y_pred)
mock_completeness_score.assert_called_once_with(labels, y_pred)
mock_v_measure_score.assert_called_once_with(labels, y_pred)
mock_adjusted_rand_score.assert_called_once_with(labels, y_pred)
assert h**o == 0
assert compl == 1
assert v_meas == 2
assert rai == 3