def test_predict_transform_copy():
    # check that the "copy" keyword works
    d = load_linnerud()
    X = d.data
    Y = d.target
    clf = pls_.PLSCanonical()
    X_copy = X.copy()
    Y_copy = Y.copy()
    clf.fit(X, Y)
    # check that results are identical with copy
    assert_array_almost_equal(clf.predict(X), clf.predict(X.copy(), copy=False))
    assert_array_almost_equal(clf.transform(X), clf.transform(X.copy(), copy=False))

    # check also if passing Y
    assert_array_almost_equal(clf.transform(X, Y),
                              clf.transform(X.copy(), Y.copy(), copy=False))
    # check that copy doesn't destroy
    # we do want to check exact equality here
    assert_array_equal(X_copy, X)
    assert_array_equal(Y_copy, Y)
    # also check that mean wasn't zero before (to make sure we didn't touch it)
    assert np.all(X.mean(axis=0) != 0) 

def test_predict_transform_copy():
    # check that the "copy" keyword works
    d = load_linnerud()
    X = d.data
    Y = d.target
    clf = pls_.PLSCanonical()
    X_copy = X.copy()
    Y_copy = Y.copy()
    clf.fit(X, Y)
    # check that results are identical with copy
    assert_array_almost_equal(clf.predict(X), clf.predict(X.copy(), copy=False))
    assert_array_almost_equal(clf.transform(X), clf.transform(X.copy(), copy=False))

    # check also if passing Y
    assert_array_almost_equal(clf.transform(X, Y),
                              clf.transform(X.copy(), Y.copy(), copy=False))
    # check that copy doesn't destroy
    # we do want to check exact equality here
    assert_array_equal(X_copy, X)
    assert_array_equal(Y_copy, Y)
    # also check that mean wasn't zero before (to make sure we didn't touch it)
    assert_true(np.all(X.mean(axis=0) != 0)) 

def test_convergence_fail():
    d = load_linnerud()
    X = d.data
    Y = d.target
    pls_bynipals = pls_.PLSCanonical(n_components=X.shape[1],
                                     max_iter=2, tol=1e-10)
    assert_warns(ConvergenceWarning, pls_bynipals.fit, X, Y) 

def test_PLSSVD():
    # Let's check the PLSSVD doesn't return all possible component but just
    # the specified number
    d = load_linnerud()
    X = d.data
    Y = d.target
    n_components = 2
    for clf in [pls_.PLSSVD, pls_.PLSRegression, pls_.PLSCanonical]:
        pls = clf(n_components=n_components)
        pls.fit(X, Y)
        assert_equal(n_components, pls.y_scores_.shape[1]) 

def test_univariate_pls_regression():
    # Ensure 1d Y is correctly interpreted
    d = load_linnerud()
    X = d.data
    Y = d.target

    clf = pls_.PLSRegression()
    # Compare 1d to column vector
    model1 = clf.fit(X, Y[:, 0]).coef_
    model2 = clf.fit(X, Y[:, :1]).coef_
    assert_array_almost_equal(model1, model2) 

def test_load_linnerud():
    res = load_linnerud()
    assert_equal(res.data.shape, (20, 3))
    assert_equal(res.target.shape, (20, 3))
    assert_equal(len(res.target_names), 3)
    assert res.DESCR
    assert os.path.exists(res.data_filename)
    assert os.path.exists(res.target_filename)

    # test return_X_y option
    check_return_X_y(res, partial(load_linnerud)) 

def test_linnerud(self):
        return
        # data must be 1-dimensional

        # data = datasets.load_linnerud()
        # df = pdml.ModelFrame(data)
        # self.assertEqual(df.shape, (150, 5)) 

def test_PLSSVD():
    # Let's check the PLSSVD doesn't return all possible component but just
    # the specified number
    d = load_linnerud()
    X = d.data
    Y = d.target
    n_components = 2
    for clf in [pls_.PLSSVD, pls_.PLSRegression, pls_.PLSCanonical]:
        pls = clf(n_components=n_components)
        pls.fit(X, Y)
        assert_equal(n_components, pls.y_scores_.shape[1]) 

def test_univariate_pls_regression():
    # Ensure 1d Y is correctly interpreted
    d = load_linnerud()
    X = d.data
    Y = d.target

    clf = pls_.PLSRegression()
    # Compare 1d to column vector
    model1 = clf.fit(X, Y[:, 0]).coef_
    model2 = clf.fit(X, Y[:, :1]).coef_
    assert_array_almost_equal(model1, model2) 

def test_pls_errors():
    d = load_linnerud()
    X = d.data
    Y = d.target
    for clf in [pls_.PLSCanonical(), pls_.PLSRegression(),
                pls_.PLSSVD()]:
        clf.n_components = 4
        assert_raise_message(ValueError, "Invalid number of components",
                             clf.fit, X, Y) 

def test_scale_and_stability():
    # We test scale=True parameter
    # This allows to check numerical stability over platforms as well

    d = load_linnerud()
    X1 = d.data
    Y1 = d.target
    # causes X[:, -1].std() to be zero
    X1[:, -1] = 1.0

    # From bug #2821
    # Test with X2, T2 s.t. clf.x_score[:, 1] == 0, clf.y_score[:, 1] == 0
    # This test robustness of algorithm when dealing with value close to 0
    X2 = np.array([[0., 0., 1.],
                   [1., 0., 0.],
                   [2., 2., 2.],
                   [3., 5., 4.]])
    Y2 = np.array([[0.1, -0.2],
                   [0.9, 1.1],
                   [6.2, 5.9],
                   [11.9, 12.3]])

    for (X, Y) in [(X1, Y1), (X2, Y2)]:
        X_std = X.std(axis=0, ddof=1)
        X_std[X_std == 0] = 1
        Y_std = Y.std(axis=0, ddof=1)
        Y_std[Y_std == 0] = 1

        X_s = (X - X.mean(axis=0)) / X_std
        Y_s = (Y - Y.mean(axis=0)) / Y_std

        for clf in [CCA(), pls_.PLSCanonical(), pls_.PLSRegression(),
                    pls_.PLSSVD()]:
            clf.set_params(scale=True)
            X_score, Y_score = clf.fit_transform(X, Y)
            clf.set_params(scale=False)
            X_s_score, Y_s_score = clf.fit_transform(X_s, Y_s)
            assert_array_almost_equal(X_s_score, X_score, decimal=4)
            assert_array_almost_equal(Y_s_score, Y_score, decimal=4)
            # Scaling should be idempotent
            clf.set_params(scale=True)
            X_score, Y_score = clf.fit_transform(X_s, Y_s)
            assert_array_almost_equal(X_s_score, X_score, decimal=4)
            assert_array_almost_equal(Y_s_score, Y_score, decimal=4) 

def test_scale_and_stability():
    # We test scale=True parameter
    # This allows to check numerical stability over platforms as well

    d = load_linnerud()
    X1 = d.data
    Y1 = d.target
    # causes X[:, -1].std() to be zero
    X1[:, -1] = 1.0

    # From bug #2821
    # Test with X2, T2 s.t. clf.x_score[:, 1] == 0, clf.y_score[:, 1] == 0
    # This test robustness of algorithm when dealing with value close to 0
    X2 = np.array([[0., 0., 1.],
                   [1., 0., 0.],
                   [2., 2., 2.],
                   [3., 5., 4.]])
    Y2 = np.array([[0.1, -0.2],
                   [0.9, 1.1],
                   [6.2, 5.9],
                   [11.9, 12.3]])

    for (X, Y) in [(X1, Y1), (X2, Y2)]:
        X_std = X.std(axis=0, ddof=1)
        X_std[X_std == 0] = 1
        Y_std = Y.std(axis=0, ddof=1)
        Y_std[Y_std == 0] = 1

        X_s = (X - X.mean(axis=0)) / X_std
        Y_s = (Y - Y.mean(axis=0)) / Y_std

        for clf in [CCA(), pls_.PLSCanonical(), pls_.PLSRegression(),
                    pls_.PLSSVD()]:
            clf.set_params(scale=True)
            X_score, Y_score = clf.fit_transform(X, Y)
            clf.set_params(scale=False)
            X_s_score, Y_s_score = clf.fit_transform(X_s, Y_s)
            assert_array_almost_equal(X_s_score, X_score)
            assert_array_almost_equal(Y_s_score, Y_score)
            # Scaling should be idempotent
            clf.set_params(scale=True)
            X_score, Y_score = clf.fit_transform(X_s, Y_s)
            assert_array_almost_equal(X_s_score, X_score)
            assert_array_almost_equal(Y_s_score, Y_score)