MAINT first step towards sklearn==0.17.1

zhangweijiqn · Apr 7, 2016 · db611bd · db611bd
1 parent 22b925f
commit db611bd
Show file tree

Hide file tree

Showing 25 changed files with 49 additions and 50 deletions.
diff --git a/.travis.yml b/.travis.yml
@@ -30,8 +30,8 @@ env:
   - TEST_DIR=/tmp/test_dir/
   - MODULE=autosklearn
   matrix:
-  - DISTRIB="conda" PYTHON_VERSION="2.7" NUMPY_VERSION="1.9" SCIPY_VERSION="0.17"
-  - DISTRIB="conda" PYTHON_VERSION="3.4" COVERAGE="true" NUMPY_VERSION="1.9" SCIPY_VERSION="0.17"
+  - DISTRIB="conda" PYTHON_VERSION="2.7"
+  - DISTRIB="conda" PYTHON_VERSION="3.4" COVERAGE="true"
 
 install:
   # Necessary for random forest

diff --git a/autosklearn/pipeline/components/classification/lda.py b/autosklearn/pipeline/components/classification/lda.py
@@ -19,7 +19,7 @@ def __init__(self, shrinkage, n_components, tol, shrinkage_factor=0.5,
         self.estimator = None
 
     def fit(self, X, Y):
-        import sklearn.lda
+        import sklearn.discriminant_analysis
         import sklearn.multiclass
 
         if self.shrinkage == "None":
@@ -36,10 +36,9 @@ def fit(self, X, Y):
         self.n_components = int(self.n_components)
         self.tol = float(self.tol)
 
-        estimator = sklearn.lda.LDA(n_components=self.n_components,
-                                    shrinkage=self.shrinkage,
-                                    tol=self.tol,
-                                    solver=solver)
+        estimator = sklearn.discriminant_analysis.LinearDiscriminantAnalysis(
+            n_components=self.n_components, shrinkage=self.shrinkage,
+            tol=self.tol, solver=solver)
 
         if len(Y.shape) == 2 and Y.shape[1] > 1:
             self.estimator = sklearn.multiclass.OneVsRestClassifier(estimator, n_jobs=1)

diff --git a/autosklearn/pipeline/components/classification/qda.py b/autosklearn/pipeline/components/classification/qda.py
@@ -16,9 +16,9 @@ def __init__(self, reg_param, random_state=None):
         self.estimator = None
 
     def fit(self, X, Y):
-        import sklearn.qda
+        import sklearn.discriminant_analysis
 
-        estimator = sklearn.qda.QDA(self.reg_param)
+        estimator = sklearn.discriminant_analysis.QuadraticDiscriminantAnalysis(self.reg_param)
 
         if len(Y.shape) == 2 and Y.shape[1] > 1:
             self.estimator = sklearn.multiclass.OneVsRestClassifier(estimator, n_jobs=1)

diff --git a/autosklearn/pipeline/components/regression/adaboost.py b/autosklearn/pipeline/components/regression/adaboost.py
@@ -25,7 +25,7 @@ def fit(self, X, Y):
         self.n_estimators = int(self.n_estimators)
         self.learning_rate = float(self.learning_rate)
         self.max_depth = int(self.max_depth)
-        base_estimator = sklearn.tree.DecisionTreeClassifier(
+        base_estimator = sklearn.tree.DecisionTreeRegressor(
             max_depth=self.max_depth)
 
         self.estimator = sklearn.ensemble.AdaBoostRegressor(

diff --git a/autosklearn/pipeline/components/regression/gaussian_process.py b/autosklearn/pipeline/components/regression/gaussian_process.py
@@ -34,8 +34,8 @@ def fit(self, X, Y):
             optimizer='Welch',
             random_state=self.random_state)
         self.scaler = sklearn.preprocessing.StandardScaler(copy=True)
-        self.scaler.fit(Y)
-        Y_scaled = self.scaler.transform(Y)
+        self.scaler.fit(Y.reshape((-1, 1)))
+        Y_scaled = self.scaler.transform(Y.reshape((-1, 1))).ravel()
         self.estimator.fit(X, Y_scaled)
         return self
 

diff --git a/autosklearn/pipeline/components/regression/k_nearest_neighbors.py b/autosklearn/pipeline/components/regression/k_nearest_neighbors.py
@@ -1,6 +1,6 @@
 from ConfigSpace.configuration_space import ConfigurationSpace
 from ConfigSpace.hyperparameters import CategoricalHyperparameter, \
-    Constant, UniformIntegerHyperparameter
+    UniformIntegerHyperparameter
 
 from autosklearn.pipeline.components.base import AutoSklearnRegressionAlgorithm
 from autosklearn.pipeline.constants import *
@@ -17,7 +17,7 @@ def fit(self, X, Y):
         import sklearn.neighbors
 
         self.estimator = \
-            sklearn.neighbors.KNeighborsClassifier(
+            sklearn.neighbors.KNeighborsRegressor(
                 n_neighbors=self.n_neighbors,
                 weights=self.weights,
                 p=self.p)

diff --git a/autosklearn/pipeline/components/regression/libsvm_svr.py b/autosklearn/pipeline/components/regression/libsvm_svr.py
@@ -71,8 +71,8 @@ def fit(self, X, Y):
         )
         self.scaler = sklearn.preprocessing.StandardScaler(copy=True)
 
-        self.scaler.fit(Y)
-        Y_scaled = self.scaler.transform(Y)
+        self.scaler.fit(Y.reshape((-1, 1)))
+        Y_scaled = self.scaler.transform(Y.reshape((-1, 1))).ravel()
         self.estimator.fit(X, Y_scaled)
         return self
 

diff --git a/autosklearn/pipeline/components/regression/sgd.py b/autosklearn/pipeline/components/regression/sgd.py
@@ -71,9 +71,9 @@ def iterative_fit(self, X, y, n_iter=1, refit=False):
                                           random_state=self.random_state)
 
             self.scaler = sklearn.preprocessing.StandardScaler(copy=True)
-            self.scaler.fit(y)
+            self.scaler.fit(y.reshape((-1, 1)))
 
-        Y_scaled = self.scaler.transform(y)
+        Y_scaled = self.scaler.transform(y.reshape((-1, 1))).ravel()
 
         self.estimator.n_iter = n_iter
         self._iterations += n_iter

diff --git a/circle.yml b/circle.yml
@@ -22,7 +22,6 @@ dependencies:
     - sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-4.6 10
     - sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.9 20
     - sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-4.9 20
-    - sudo update-alternatives --install /usr/bin/x86_64-linux-gnu-gcc x86_64-linux-gnu-gcc /usr/bin/g++-4.9 20
     # from scikit-learn contrib
     - sudo -E apt-get -yq remove texlive-binaries --purge
     - sudo apt-get update

diff --git a/requ.txt b/requ.txt
@@ -6,7 +6,7 @@ nose
 numpy>=1.9.0
 scipy>=0.14.1
 
-scikit-learn==0.16.1
+scikit-learn==0.17.1
 
 lockfile
 joblib

diff --git a/setup.py b/setup.py
@@ -34,7 +34,7 @@
     packages=setuptools.find_packages(exclude=['test']),
     install_requires=['numpy>=1.9.0',
                       'scipy>=0.14.1',
-                      'scikit-learn==0.16.1',
+                      'scikit-learn==0.17.1',
                       'lockfile',
                       'psutil',
                       'pyyaml',

diff --git a/test/test_pipeline/components/classification/test_adaboost.py b/test/test_pipeline/components/classification/test_adaboost.py
@@ -22,14 +22,14 @@ def test_default_configuration_iris_predict_proba(self):
         for i in range(10):
             predictions, targets = \
                 _test_classifier_predict_proba(AdaboostClassifier)
-            self.assertAlmostEqual(0.34244204343758322,
+            self.assertAlmostEqual(0.22452300738472031,
                                    sklearn.metrics.log_loss(targets, predictions))
 
     def test_default_configuration_iris_sparse(self):
         for i in range(10):
             predictions, targets = \
                 _test_classifier(AdaboostClassifier, sparse=True)
-            self.assertAlmostEqual(0.88,
+            self.assertAlmostEqual(0.85999999999999999,
                                    sklearn.metrics.accuracy_score(targets,
                                                                   predictions))
 
@@ -39,7 +39,7 @@ def test_default_configuration_multilabel(self):
                 _test_classifier(classifier=AdaboostClassifier,
                                  dataset='digits',
                                  make_multilabel=True)
-            self.assertAlmostEqual(0.80933874118770355,
+            self.assertAlmostEqual(0.79529966660329099,
                                    sklearn.metrics.average_precision_score(
                                        targets, predictions))
 
@@ -49,7 +49,7 @@ def test_default_configuration_multilabel_predict_proba(self):
                 _test_classifier_predict_proba(classifier=AdaboostClassifier,
                                  make_multilabel=True)
             self.assertEqual(predictions.shape, ((50, 3)))
-            self.assertAlmostEqual(0.97856971820815897,
+            self.assertAlmostEqual(0.9722131915406923,
                                    sklearn.metrics.average_precision_score(
                                        targets, predictions))
 
@@ -59,7 +59,7 @@ def test_default_configuration_binary(self):
                 _test_classifier(classifier=AdaboostClassifier,
                                  dataset='digits', sparse=True,
                                  make_binary=True)
-            self.assertAlmostEqual(0.93199757134183359,
+            self.assertAlmostEqual(0.93564055859137829,
                                    sklearn.metrics.accuracy_score(
                                        targets, predictions))
 

diff --git a/test/test_pipeline/components/classification/test_gaussian_nb.py b/test/test_pipeline/components/classification/test_gaussian_nb.py
@@ -41,7 +41,7 @@ def test_default_configuration_multilabel(self):
                 _test_classifier(classifier=GaussianNB,
                                  dataset='digits',
                                  make_multilabel=True)
-            self.assertAlmostEqual(0.70602999913720499,
+            self.assertAlmostEqual(0.71507312748717466,
                                    sklearn.metrics.average_precision_score(
                                        targets, predictions))
 

diff --git a/test/test_pipeline/components/classification/test_gradient_boosting.py b/test/test_pipeline/components/classification/test_gradient_boosting.py
@@ -40,7 +40,7 @@ def test_default_configuration_multilabel(self):
                 _test_classifier(classifier=GradientBoostingClassifier,
                                  dataset='digits',
                                  make_multilabel=True)
-            self.assertAlmostEqual(0.83835860291198527,
+            self.assertAlmostEqual(0.84004577632243804,
                                    sklearn.metrics.average_precision_score(
                                        targets, predictions))
 

diff --git a/test/test_pipeline/components/classification/test_lda.py b/test/test_pipeline/components/classification/test_lda.py
@@ -5,7 +5,7 @@
 
 import numpy as np
 import sklearn.metrics
-import sklearn.lda
+import sklearn.discriminant_analysis
 
 
 class LDAComponentTest(unittest.TestCase):
@@ -53,7 +53,7 @@ def test_default_configuration_predict_proba_multilabel(self):
                                        targets, predictions))
 
     def test_target_algorithm_multioutput_multiclass_support(self):
-        cls = sklearn.lda.LDA()
+        cls = sklearn.discriminant_analysis.LinearDiscriminantAnalysis()
         X = np.random.random((10, 10))
         y = np.random.randint(0, 1, size=(10, 10))
         self.assertRaisesRegexp(ValueError, 'bad input shape \(10, 10\)',

diff --git a/test/test_pipeline/components/classification/test_qda.py b/test/test_pipeline/components/classification/test_qda.py
@@ -5,7 +5,7 @@
 
 import numpy as np
 import sklearn.metrics
-import sklearn.qda
+import sklearn.discriminant_analysis
 
 
 class QDAComponentTest(unittest.TestCase):
@@ -75,7 +75,7 @@ def test_default_configuration_predict_proba_multilabel(self):
                                        targets, predictions))
 
     def test_target_algorithm_multioutput_multiclass_support(self):
-        cls = sklearn.qda.QDA()
+        cls = sklearn.discriminant_analysis.QuadraticDiscriminantAnalysis()
         X = np.random.random((10, 10))
         y = np.random.randint(0, 1, size=(10, 10))
         self.assertRaisesRegexp(ValueError, 'bad input shape \(10, 10\)',

diff --git a/test/test_pipeline/components/data_preprocessing/test_balancing.py b/test/test_pipeline/components/data_preprocessing/test_balancing.py
@@ -67,12 +67,12 @@ def test_weighting_effect(self):
             random_state=1)
 
         for name, clf, acc_no_weighting, acc_weighting in \
-                [('adaboost', AdaboostClassifier, 0.709, 0.662),
-                 ('decision_tree', DecisionTree, 0.683, 0.726),
-                 ('extra_trees', ExtraTreesClassifier, 0.812, 0.812),
+                [('adaboost', AdaboostClassifier, 0.709, 0.658),
+                 ('decision_tree', DecisionTree, 0.683, 0.701),
+                 ('extra_trees', ExtraTreesClassifier, 0.812, 0.8),
                  ('gradient_boosting', GradientBoostingClassifier,
                     0.800, 0.760),
-                 ('random_forest', RandomForest, 0.849, 0.780),
+                 ('random_forest', RandomForest, 0.846, 0.792),
                  ('libsvm_svc', LibSVM_SVC, 0.571, 0.658),
                  ('liblinear_svc', LibLinear_SVC, 0.685, 0.699),
                  ('sgd', SGD, 0.65384615384615385, 0.38795986622073581)]:

diff --git a/test/test_pipeline/components/feature_preprocessing/test_extra_trees_regression.py b/test/test_pipeline/components/feature_preprocessing/test_extra_trees_regression.py
@@ -35,7 +35,7 @@ def test_default_configuration_regression(self):
             predictor = regressor.fit(X_train_trans, Y_train)
             predictions = predictor.predict(X_test_trans)
             accuracy = sklearn.metrics.mean_squared_error(predictions, Y_test)
-            self.assertAlmostEqual(accuracy, 19.812321348314608, places=2)
+            self.assertAlmostEqual(accuracy, 20.193400000000004, places=2)
 
     def test_default_configuration_classify_sparse(self):
         for i in range(2):
@@ -56,7 +56,7 @@ def test_default_configuration_classify_sparse(self):
             predictor = regressor.fit(X_train_trans, Y_train)
             predictions = predictor.predict(X_test_trans)
             accuracy = sklearn.metrics.mean_squared_error(predictions, Y_test)
-            self.assertAlmostEqual(accuracy, 62.741933389903252, places=2)
+            self.assertAlmostEqual(accuracy, 62.485374939528718, places=2)
 
     def test_preprocessing_dtype(self):
         super(ExtreTreesRegressionComponentTest, self).\

diff --git a/test/test_pipeline/components/feature_preprocessing/test_feature_agglomeration.py b/test/test_pipeline/components/feature_preprocessing/test_feature_agglomeration.py
@@ -31,7 +31,7 @@ def test_default_configuration_classify(self):
             predictor = classifier.fit(X_train_trans, Y_train)
             predictions = predictor.predict(X_test_trans)
             accuracy = sklearn.metrics.accuracy_score(predictions, Y_test)
-            self.assertAlmostEqual(accuracy, 0.8026715)
+            self.assertAlmostEqual(accuracy, 0.81056466302367947)
 
     def test_preprocessing_dtype(self):
         super(FeatureAgglomerationComponentTest,

diff --git a/test/test_pipeline/components/regression/test_adaboost.py b/test/test_pipeline/components/regression/test_adaboost.py
@@ -12,14 +12,14 @@ def test_default_configuration(self):
         for i in range(10):
             predictions, targets = \
                 _test_regressor(AdaboostRegressor, dataset='boston')
-            self.assertAlmostEqual(0.11053868761882502,
+            self.assertAlmostEqual(0.53163164558327014,
                                    sklearn.metrics.r2_score(targets,
                                                             predictions))
 
     def test_default_configuration_sparse(self):
         for i in range(10):
             predictions, targets = \
                 _test_regressor(AdaboostRegressor, sparse=True, dataset='boston')
-            self.assertAlmostEqual(-0.077540100211211049,
+            self.assertAlmostEqual(0.22163148231037866,
                                    sklearn.metrics.r2_score(targets,
                                                             predictions))
diff --git a/test/test_pipeline/components/regression/test_extra_trees.py b/test/test_pipeline/components/regression/test_extra_trees.py
@@ -12,22 +12,22 @@ def test_default_configuration(self):
         for i in range(10):
             predictions, targets = \
                 _test_regressor(ExtraTreesRegressor)
-            self.assertAlmostEqual(0.4269923975466271,
+            self.assertAlmostEqual(0.42779639951661386,
                                    sklearn.metrics.r2_score(targets,
                                                              predictions))
 
     def test_default_configuration_sparse(self):
         for i in range(10):
             predictions, targets = \
                 _test_regressor(ExtraTreesRegressor, sparse=True)
-            self.assertAlmostEqual(0.26287621251507987,
+            self.assertAlmostEqual(0.26270282847272175,
                                    sklearn.metrics.r2_score(targets,
                                                             predictions))
 
     def test_default_configuration_iterative_fit(self):
         for i in range(10):
             predictions, targets = \
                 _test_regressor_iterative_fit(ExtraTreesRegressor)
-            self.assertAlmostEqual(0.4269923975466271,
+            self.assertAlmostEqual(0.42779639951661386,
                                    sklearn.metrics.r2_score(targets,
                                                             predictions))
diff --git a/test/test_pipeline/components/regression/test_gaussian_process.py b/test/test_pipeline/components/regression/test_gaussian_process.py
@@ -8,7 +8,8 @@
 
 class GaussianProcessComponentTest(unittest.TestCase):
     def test_default_configuration(self):
-        for i in range(10):
+        # Only twice to reduce the number of warning printed to the command line
+        for i in range(2):
             # Float32 leads to numeric instabilities
             predictions, targets = _test_regressor(GaussianProcess,
                                                    dataset='boston')

diff --git a/...ne/components/regression/liblinear_svr.py → ...mponents/regression/test_liblinear_svr.py b/...ne/components/regression/liblinear_svr.py → ...mponents/regression/test_liblinear_svr.py
@@ -12,6 +12,6 @@ def test_default_configuration(self):
         for i in range(10):
             predictions, targets = _test_regressor(LibLinear_SVR,
                                                    dataset='boston')
-            self.assertAlmostEqual(0.54372712745256768,
+            self.assertAlmostEqual(0.67714499792824023,
                                    sklearn.metrics.r2_score(y_true=targets,
                                                             y_pred=predictions))
diff --git a/test/test_pipeline/components/regression/test_random_forests.py b/test/test_pipeline/components/regression/test_random_forests.py
@@ -11,19 +11,19 @@ def test_default_configuration(self):
         for i in range(10):
 
             predictions, targets = _test_regressor(RandomForest)
-            self.assertAlmostEqual(0.41224692924630502,
+            self.assertAlmostEqual(0.40965687834764064,
                 sklearn.metrics.r2_score(y_true=targets, y_pred=predictions))
 
 
     def test_default_configuration_sparse(self):
         for i in range(10):
             predictions, targets = _test_regressor(RandomForest, sparse=True)
-            self.assertAlmostEqual(0.24117530425422551,
+            self.assertAlmostEqual(0.24147637508106434,
                 sklearn.metrics.r2_score(y_true=targets, y_pred=predictions))
 
     def test_default_configuration_iterative_fit(self):
         for i in range(10):
             predictions, targets = \
                 _test_regressor_iterative_fit(RandomForest)
-            self.assertAlmostEqual(0.41224692924630502,
+            self.assertAlmostEqual(0.40965687834764064,
                 sklearn.metrics.r2_score(y_true=targets, y_pred=predictions))
diff --git a/test/test_pipeline/test_regression.py b/test/test_pipeline/test_regression.py
@@ -263,7 +263,7 @@ def test_default_configuration(self):
             predictions = auto.predict(copy.deepcopy(X_test))
             # The lower the worse
             r2_score = sklearn.metrics.r2_score(Y_test, predictions)
-            self.assertAlmostEqual(0.41626416529791199, r2_score)
+            self.assertAlmostEqual(0.41732302035060087, r2_score)
             model_score = auto.score(copy.deepcopy(X_test), Y_test)
             self.assertEqual(model_score, r2_score)