FIX: linnerud dataset mixed variables.

Variables exercise and physiological and should be the other way
round.

Thanks to James Bergstra for reporting the bug.

scikits/learn/datasets/data/linnerud_exercise.csv

-Weight Waist Pulse
-191 36 50
-189 37 52
-193 38 58
-162 35 62
-189 35 46
-182 36 56
-211 38 56
-167 34 60
-176 31 74
-154 33 56
-169 34 50
-166 33 52
-154 34 64
-247 46 50
-193 36 46
-202 37 62
-176 37 54
-157 32 52
-156 33 54
-138 33 68
+Chins Situps Jumps
+5 162 60
+2 110 60
+12 101 101
+12 105 37
+13 155 58
+4 101 42
+8 101 38
+6 125 40
+15 200 40
+17 251 250
+17 120 38
+13 210 115
+14 215 105
+1 50 50
+6 70 31
+12 210 120
+4 60 25
+11 230 80
+15 225 73
+2 110 43

scikits/learn/datasets/data/linnerud_physiological.csv

-Chins Situps Jumps
-5 162 60
-2 110 60
-12 101 101
-12 105 37
-13 155 58
-4 101 42
-8 101 38
-6 125 40
-15 200 40
-17 251 250
-17 120 38
-13 210 115
-14 215 105
-1 50 50
-6 70 31
-12 210 120
-4 60 25
-11 230 80
-15 225 73
-2 110 43
+Weight Waist Pulse
+191 36 50
+189 37 52
+193 38 58
+162 35 62
+189 35 46
+182 36 56
+211 38 56
+167 34 60
+176 31 74
+154 33 56
+169 34 50
+166 33 52
+154 34 64
+247 46 50
+193 36 46
+202 37 62
+176 37 54
+157 32 52
+156 33 54
+138 33 68

scikits/learn/tests/test_pls.py

@@ -86,27 +86,27 @@ def test_pls():
 
     pls_ca = pls_bynipals
     x_loadings = np.array(
-          [[-0.66591531,  0.77356014],
-           [-0.67602366, -0.62873035],
-           [ 0.35892139, -0.11993352]])
+       [[-0.61470416,  0.37877695],
+        [-0.65625755,  0.01196893],
+        [-0.51733059, -0.93984954]])
     assert_array_almost_equal(pls_ca.x_loadings_, x_loadings)
 
     y_loadings = np.array(
-          [[ 0.6147046 ,  0.37877736],
-           [ 0.65625787,  0.01196607],
-           [ 0.51733043, -0.93984895]])
+        [[ 0.66591533,  0.77358148],
+         [ 0.67602364, -0.62871191],
+         [-0.35892128, -0.11981924]])
     assert_array_almost_equal(pls_ca.y_loadings_, y_loadings)
 
     x_weights = np.array(
-          [[-0.58989082,  0.78900159],
-           [-0.77134081, -0.61352087],
-           [ 0.23887693, -0.03269003]])
+        [[-0.61330704,  0.25616119],
+         [-0.74697144,  0.11930791],
+         [-0.25668686, -0.95924297]])
     assert_array_almost_equal(pls_ca.x_weights_, x_weights)
 
     y_weights = np.array(
-          [[ 0.61330742,  0.25616374],
-           [ 0.74697171,  0.11930342],
-           [ 0.25668516, -0.95924284]])
+        [[ 0.58989127,  0.7890047 ],
+         [ 0.77134053, -0.61351791],
+         [-0.2388767 , -0.03267062]])
     assert_array_almost_equal(pls_ca.y_weights_, y_weights)
 
     # 2) Regression PLS (PLS2): "Non regression test"
@@ -115,49 +115,31 @@ def test_pls():
     pls2.fit(X, Y)
 
     x_loadings = np.array(
-          [[-0.66591531, -0.01976472],
-           [-0.67602366, -0.35471175],
-           [ 0.35892139, -1.19418263]])
+        [[-0.61470416, -0.24574278],
+         [-0.65625755, -0.14396183],
+         [-0.51733059,  1.00609417]])
     assert_array_almost_equal(pls2.x_loadings_, x_loadings)
 
     y_loadings = np.array(
-          [[ 0.34163079,  0.33638156],
-           [ 0.41608584,  0.29078321],
-           [ 0.1429814 ,  0.0651935 ]])
+        [[ 0.32456184,  0.29892183],
+         [ 0.42439636,  0.61970543],
+         [-0.13143144, -0.26348971]])
     assert_array_almost_equal(pls2.y_loadings_, y_loadings)
 
     x_weights = np.array(
-          [[-0.58989082,  0.46882134],
-           [-0.77134081, -0.56802033],
-           [ 0.23887693, -0.67643141]])
+        [[-0.61330704, -0.00443647],
+         [-0.74697144, -0.32172099],
+         [-0.25668686,  0.94682413]])
     assert_array_almost_equal(pls2.x_weights_, x_weights)
 
     y_weights = np.array(
-          [[ 0.61330742,  0.74851767],
-           [ 0.74697171,  0.64705202],
-           [ 0.25668516,  0.14506885]])
+        [[ 0.58989127,  0.40572461],
+         [ 0.77134053,  0.84112205],
+         [-0.2388767 , -0.35763282]])
     assert_array_almost_equal(pls2.y_weights_, y_weights)
 
-    ypred = np.array(
-           [[ 9.34053037e+00,   1.39572471e+02,   6.75634346e+01],
-           [  8.35625179e+00,   1.28543567e+02,   6.52136587e+01],
-           [  6.88442847e+00,   1.12444632e+02,   6.19046849e+01],
-           [  9.48402047e+00,   1.51099673e+02,   7.30739923e+01],
-           [  1.05852278e+01,   1.53487226e+02,   7.05181552e+01],
-           [  8.90885030e+00,   1.38282450e+02,   6.83808597e+01],
-           [  6.77862431e+00,   1.07612118e+02,   5.97520250e+01],
-           [  1.04183596e+01,   1.61076073e+02,   7.50477312e+01],
-           [  1.13829272e+01,   1.78285171e+02,   8.06854024e+01],
-           [  1.18461685e+01,   1.78850010e+02,   7.93812540e+01],
-           [  1.13542309e+01,   1.67786960e+02,   7.53148497e+01],
-           [  1.20340762e+01,   1.77605016e+02,   7.80842300e+01],
-           [  1.02470983e+01,   1.62671902e+02,   7.64701009e+01],
-           [  1.70939023e-01,   1.74451012e+01,   3.55750249e+01],
-           [  9.69490309e+00,   1.41835825e+02,   6.75198672e+01],
-           [  7.17068297e+00,   1.16570160e+02,   6.30663554e+01],
-           [  8.37882344e+00,   1.31621907e+02,   6.67395903e+01],
-           [  1.30076335e+01,   1.91010591e+02,   8.17092462e+01],
-           [  1.20067084e+01,   1.79630851e+02,   7.92341246e+01],
-           [  1.09495154e+01,   1.75568297e+02,   8.07654128e+01]])
-
-    assert_array_almost_equal(pls2.predict(X), ypred)
+    ypred_2 = np.array(
+        [[ 180.33278555,   35.57034871,   56.06817703],
+         [ 192.06235219,   37.95306771,   54.12925192]])
+
+    assert_array_almost_equal(pls2.predict(X[:2]), ypred_2)