def test_GM_fit_lowiterations():
X, y = data.categorical_2Dmatrix_data_big()
np.random.seed(0)
GM = gaussianmixture.GaussianMixture()
GM.fit(X, iterations=1)
assert GM.learned
np.testing.assert_array_almost_equal(GM.mus[0], [2.22, 2.5], decimal=1)
def test_LVQ():
X, y = data.categorical_2Dmatrix_data_big()
lvq = prototypemthods.LearningVectorQuantization()
lvq.fit(X, y, n_prototypes=3)
assert [0, 1] == sorted(lvq.prototypes.keys())
assert (3, 2) == lvq.prototypes[0].shape
assert (3, 2) == lvq.prototypes[1].shape
def test_test_train_splitter():
X, y = data.categorical_2Dmatrix_data_big()
X_train, X_test, y_train, y_test = modelselection.test_train_splitter(X, y)
assert X_train.shape == (9, 2)
assert X_test.shape == (2, 2)
assert len(y_train) == 9
assert len(y_test) == 2
def test_Perceptron():
X, y = data.categorical_2Dmatrix_data_big()
y = (y * 2) - 1
perceptron = svm.Perceptron()
perceptron.fit(X, y)
predictions = perceptron.predict(X)
assert (predictions == y).sum() > 7
def test_test_train_splitter():
X, y = data.categorical_2Dmatrix_data_big()
X_train, X_test, y_train, y_test = modelselection.test_train_splitter(X, y)
assert X_train.shape == (9, 2)
assert X_test.shape == (2, 2)
assert len(y_train) == 9
assert len(y_test) == 2
def test_KMediods():
X, y = data.categorical_2Dmatrix_data_big()
km = prototypemthods.KMediods()
km.fit(X)
assignments = km.sample_assignments
reversed_assignments = (assignments - 1) * -1
assert np.array_equal(assignments, y) or \
np.array_equal(reversed_assignments, y)
def test_LVQ_prediction():
X, y = data.categorical_2Dmatrix_data_big()
lvq = prototypemthods.LearningVectorQuantization()
lvq.fit(X, y, n_prototypes=3)
prediction = lvq.predict(X[0])
assert prediction == y[0]
prediction = lvq.predict(X[-1])
assert prediction == y[-1]
def test_DANN_prediction():
X, y = data.categorical_2Dmatrix_data_big()
dann = prototypemthods.DANN()
dann.fit(X, y, neighborhood_size=3)
prediction = dann.predict(X[0])
assert prediction == y[0]
prediction = dann.predict(X[-1])
assert prediction == y[-1]
def test_KMediods_prediction():
X, y = data.categorical_2Dmatrix_data_big()
km = prototypemthods.KMediods()
km.fit(X)
assignments = km.sample_assignments
reversed_assignments = (assignments - 1) * -1
prediction = km.predict(X[0])
np.testing.assert_array_almost_equal(prediction, [3.0, 3.0])
def test_GM_predict_probs():
X, y = data.categorical_2Dmatrix_data_big()
np.random.seed(0)
GM = gaussianmixture.GaussianMixture()
GM.fit(X)
max_class, class_probs = GM.predict(X[0], probs=True)
assert max_class == 0
assert np.isclose(class_probs[0], 1)
assert np.isclose(class_probs[1], 0)
def test_GM_predict():
X, y = data.categorical_2Dmatrix_data_big()
np.random.seed(0)
GM = gaussianmixture.GaussianMixture()
GM.fit(X)
pred_one = GM.predict(X[0])
pred_two = GM.predict(X[-1])
assert pred_one == y[0]
assert pred_two == y[-1]
def test_GM_fit_lowiterations():
X, y = data.categorical_2Dmatrix_data_big()
np.random.seed(0)
GM = gaussianmixture.GaussianMixture()
GM.fit(X, iterations=1)
assert GM.learned
np.testing.assert_array_almost_equal(GM.mus[0],
[2.22, 2.5],
decimal=1)
def test_SupportVectorMachine():
X, y = data.categorical_2Dmatrix_data_big()
y = (y * 2) - 1
SVM = svm.SupportVectorMachine()
SVM.fit(X, y)
prediction = SVM.predict(X[0])
assert prediction == y[0]
prediction = SVM.predict(X[-1])
assert prediction == y[-1]
def test_GM_predict_probs():
X, y = data.categorical_2Dmatrix_data_big()
np.random.seed(0)
GM = gaussianmixture.GaussianMixture()
GM.fit(X)
max_class, class_probs = GM.predict(X[0], probs=True)
assert max_class == 0
assert np.isclose(class_probs[0], 1)
assert np.isclose(class_probs[1], 0)
def test_GM_predict():
X, y = data.categorical_2Dmatrix_data_big()
np.random.seed(0)
GM = gaussianmixture.GaussianMixture()
GM.fit(X)
pred_one = GM.predict(X[0])
pred_two = GM.predict(X[-1])
assert pred_one == y[0]
assert pred_two == y[-1]
def test_RegressionTree():
tree = treemethods.RegressionTree()
X, y = data.categorical_2Dmatrix_data_big()
tree.fit(X, y, 3)
assert tree.predict(X[0]) == y[0]
assert tree.predict(X[-1]) == y[-1]
def test_PrimRegression():
tree = treemethods.PrimRegression()
X, y = data.categorical_2Dmatrix_data_big()
tree.fit(X, y, 1)
assert tree.predict(X[1]) == y[1]
assert np.isclose(tree.predict(X[-1]), 0.6666, 1)
def test_DiscreteAdaBoost():
tree = treemethods.DiscreteAdaBoost()
X, y = data.categorical_2Dmatrix_data_big()
tree.fit(X, y, 3)
assert tree.predict(X[0]) == y[0]
assert tree.predict(X[-1]) == y[-1]
def test_DANN():
X, y = data.categorical_2Dmatrix_data_big()
dann = prototypemthods.DANN()
dann.fit(X, y)
assert dann.learned
def test_GradientBoostingRegression():
tree = treemethods.GradientBoostingRegression()
X, y = data.categorical_2Dmatrix_data_big()
tree.fit(X, y, 3)
assert np.isclose(tree.predict(X[0]), 0.3976, 1)
assert np.isclose(tree.predict(X[-1]), y[-1], 1)
def test_knearestneighbor_regression():
X, y = data.categorical_2Dmatrix_data_big()
knn = prototypemthods.KNearestNeighbor()
knn.fit(X, y)
prediction = knn.predict(X[0])
assert prediction == y[0]
