def test_polynomial_features_slow(self):
X1 = numpy.arange(6)[:, numpy.newaxis]
P1 = numpy.hstack([numpy.ones_like(X1), X1, X1**2, X1**3])
deg1 = 3
X2 = numpy.arange(6).reshape((3, 2))
x1 = X2[:, :1]
x2 = X2[:, 1:]
P2 = numpy.hstack([
x1**0 * x2**0, x1**1 * x2**0, x1**0 * x2**1, x1**2 * x2**0,
x1**1 * x2**1, x1**0 * x2**2
])
deg2 = 2
for (deg, X, P) in [(deg1, X1, P1), (deg2, X2, P2)]:
poly = PolynomialFeatures(deg, include_bias=True)
P_test = poly.fit_transform(X)
self.assertEqual(P_test, P)
names = poly.get_feature_names()
ext = ExtendedFeatures(kind='poly-slow', poly_degree=deg)
e_test = ext.fit_transform(X)
e_names = ext.get_feature_names()
self.assertEqual(len(names), len(e_names))
self.assertEqual(names, e_names)
self.assertEqual(P_test, P)
self.assertEqual(P_test.shape, e_test.shape)
self.assertEqual(P_test, e_test)
def fcts_model(X, y):
model1 = SGDClassifier()
model2 = make_pipeline(PolynomialFeatures(), SGDClassifier())
model3 = make_pipeline(ExtendedFeatures(kind='poly'), SGDClassifier())
model4 = make_pipeline(ExtendedFeatures(kind='poly-slow'), SGDClassifier())
model1.fit(PolynomialFeatures().fit_transform(X), y)
model2.fit(X, y)
model3.fit(X, y)
model4.fit(X, y)
def partial_fit_model1(X, y, model=model1):
return model.partial_fit(X, y)
def partial_fit_model2(X, y, model=model2):
X2 = model.steps[0][1].transform(X)
return model.steps[1][1].partial_fit(X2, y)
def partial_fit_model3(X, y, model=model3):
X2 = model.steps[0][1].transform(X)
return model.steps[1][1].partial_fit(X2, y)
def partial_fit_model4(X, y, model=model4):
X2 = model.steps[0][1].transform(X)
return model.steps[1][1].partial_fit(X2, y)
return (partial_fit_model1, partial_fit_model2, partial_fit_model3,
partial_fit_model4)
def test_polynomial_features_bigger_ionly(self):
X = numpy.arange(30).reshape((5, 6))
for deg in (1, 2, 3, 4, 5):
poly = PolynomialFeatures(deg,
include_bias=True,
interaction_only=True)
X_sk = poly.fit_transform(X)
names_sk = poly.get_feature_names()
ext = ExtendedFeatures(poly_degree=deg,
poly_include_bias=True,
poly_interaction_only=True)
X_ext = ext.fit_transform(X)
inames = ["x%d" % i for i in range(0, X.shape[1])]
names_ext = ext.get_feature_names(inames)
self.assertEqual(len(names_sk), len(names_ext))
self.assertEqual(names_sk, names_ext)
names_ext = ext.get_feature_names()
self.assertEqual(len(names_sk), len(names_ext))
self.assertEqual(names_sk, names_ext)
self.assertEqual(X_sk.shape, X_ext.shape)
self.assertEqual(X_sk, X_ext)
def __init__(self, dim=None, **opts):
# Models are fitted here. Every not measured
# should take place here.
assert dim is not None
BenchPerfTest.__init__(self, **opts)
self.model1 = SGDClassifier()
self.model2 = make_pipeline(PolynomialFeatures(), SGDClassifier())
self.model3 = make_pipeline(
ExtendedFeatures(kind='poly'), SGDClassifier())
self.model4 = make_pipeline(ExtendedFeatures(
kind='poly-slow'), SGDClassifier())
X, y = random_binary_classification(10000, dim)
self.model1.fit(PolynomialFeatures().fit_transform(X), y)
self.model2.fit(X, y)
self.model3.fit(X, y)
self.model4.fit(X, y)
def test_polynomial_features_nobias_ionly_slow(self):
X1 = numpy.arange(6)[:, numpy.newaxis]
P1 = numpy.hstack([numpy.ones_like(X1), X1, X1**2, X1**3])
deg1 = 3
X2 = numpy.arange(6).reshape((3, 2))
x1 = X2[:, :1]
x2 = X2[:, 1:]
P2 = numpy.hstack([
x1**0 * x2**0, x1**1 * x2**0, x1**0 * x2**1, x1**2 * x2**0,
x1**1 * x2**1, x1**0 * x2**2
])
deg2 = 2
for (deg, X, P) in [(deg1, X1, P1), (deg2, X2, P2)]:
fc = [1] if deg == 3 else [1, 2, 4]
poly = PolynomialFeatures(deg,
include_bias=False,
interaction_only=True)
P_test = poly.fit_transform(X)
names = poly.get_feature_names()
self.assertEqual(P_test, P[:, fc])
ext = ExtendedFeatures(kind="poly-slow",
poly_degree=deg,
poly_include_bias=False,
poly_interaction_only=True)
e_test = ext.fit_transform(X)
e_names = ext.get_feature_names()
self.assertEqual(len(names), len(e_names))
self.assertEqual(names, e_names)
self.assertEqual(P_test, P[:, fc])
self.assertEqual(P_test.shape, e_test.shape)
self.assertEqual(P_test, e_test)
def test_polynomial_features_sparse(self):
dtype = numpy.float64
rng = numpy.random.RandomState(0) # pylint: disable=E1101
X = rng.randint(0, 2, (100, 2))
X_sparse = sparse.csr_matrix(X)
est = PolynomialFeatures(2)
Xt_sparse = est.fit_transform(X_sparse.astype(dtype))
Xt_dense = est.fit_transform(X.astype(dtype))
self.assertIsInstance(Xt_sparse,
(sparse.csc_matrix, sparse.csr_matrix))
self.assertEqual(Xt_sparse.dtype, Xt_dense.dtype)
self.assertEqual(Xt_sparse.A, Xt_dense)
est = ExtendedFeatures(poly_degree=2)
Xt_sparse = est.fit_transform(X_sparse.astype(dtype))
Xt_dense = est.fit_transform(X.astype(dtype))
self.assertIsInstance(Xt_sparse, sparse.csc_matrix)
self.assertEqual(Xt_sparse.dtype, Xt_dense.dtype)
self.assertEqual(Xt_sparse.A, Xt_dense)
def polynomial_features_csr_X_zero_row(self, zero_row_index, deg,
interaction_only):
X_csr = sparse_random(3, 10, 1.0, random_state=0).tocsr()
X_csr[zero_row_index, :] = 0.0
X = X_csr.toarray()
est = ExtendedFeatures(poly_degree=deg,
poly_include_bias=False,
poly_interaction_only=interaction_only)
est.fit(X)
poly = PolynomialFeatures(degree=deg,
include_bias=False,
interaction_only=interaction_only)
poly.fit(X)
self.assertEqual(poly.get_feature_names(), est.get_feature_names())
Xt_dense1 = est.fit_transform(X)
Xt_dense2 = poly.fit_transform(X)
self.assertEqual(Xt_dense1, Xt_dense2)