scikit-learn · ogrisel · Feb 20, 2017 · Jan 30, 2017 · Feb 14, 2017 · Feb 20, 2017
diff --git a/.travis.yml b/.travis.yml
@@ -36,6 +36,11 @@ env:
     - DISTRIB="conda" PYTHON_VERSION="3.6" INSTALL_MKL="true"
       NUMPY_VERSION="1.11.2" SCIPY_VERSION="0.18.1" PANDAS_VERSION="0.19.1"
       CYTHON_VERSION="0.25.2"
+    # This environment use pytest to run the tests. It uses the newest
+    # supported anaconda env. It also runs tests requiring Pandas.
+    - USE_PYTEST="true" DISTRIB="conda" PYTHON_VERSION="3.6" INSTALL_MKL="true"
+      NUMPY_VERSION="1.11.2" SCIPY_VERSION="0.18.1" PANDAS_VERSION="0.19.1"
+      CYTHON_VERSION="0.25.2"
     # flake8 linting on diff wrt common ancestor with upstream/master
     - RUN_FLAKE8="true" SKIP_TESTS="true"
       DISTRIB="conda" PYTHON_VERSION="3.5" INSTALL_MKL="true"

diff --git a/build_tools/travis/install.sh b/build_tools/travis/install.sh
@@ -51,13 +51,13 @@ if [[ "$DISTRIB" == "conda" ]]; then
     # Configure the conda environment and put it in the path using the
     # provided versions
     if [[ "$INSTALL_MKL" == "true" ]]; then
-        conda create -n testenv --yes python=$PYTHON_VERSION pip nose \
+        conda create -n testenv --yes python=$PYTHON_VERSION pip nose pytest \
             numpy=$NUMPY_VERSION scipy=$SCIPY_VERSION \
             mkl cython=$CYTHON_VERSION \
             ${PANDAS_VERSION+pandas=$PANDAS_VERSION}
 
     else
-        conda create -n testenv --yes python=$PYTHON_VERSION pip nose \
+        conda create -n testenv --yes python=$PYTHON_VERSION pip nose pytest \
             numpy=$NUMPY_VERSION scipy=$SCIPY_VERSION \
             nomkl cython=$CYTHON_VERSION \
             ${PANDAS_VERSION+pandas=$PANDAS_VERSION}

diff --git a/build_tools/travis/test_script.sh b/build_tools/travis/test_script.sh
@@ -21,6 +21,11 @@ except ImportError:
 python -c "import multiprocessing as mp; print('%d CPUs' % mp.cpu_count())"
 
 run_tests() {
+    if [[ "$USE_PYTEST" == "true" ]]; then
+        TEST_CMD="pytest --showlocals --pyargs"
+    else
+        TEST_CMD="nosetests --with-coverage" # --with-timer --timer-top-n 20"
+    fi
     # Get into a temp directory to run test from the installed scikit learn and
     # check if we do not leave artifacts
     mkdir -p $TEST_DIR
@@ -34,10 +39,9 @@ run_tests() {
     export SKLEARN_SKIP_NETWORK_TESTS=1
 
     if [[ "$COVERAGE" == "true" ]]; then
-        nosetests -s --with-coverage --with-timer --timer-top-n 20 sklearn
-    else
-        nosetests -s --with-timer --timer-top-n 20 sklearn
+        TEST_CMD="$TEST_CMD --with-coverage"
     fi
+    $TEST_CMD sklearn
 
     # Test doc
     cd $OLDPWD

diff --git a/setup.cfg b/setup.cfg
@@ -22,6 +22,13 @@ doctest-fixtures = _fixture
 ignore-files=^setup\.py$
 #doctest-options = +ELLIPSIS,+NORMALIZE_WHITESPACE
 
+[tool:pytest]
+# disable-pytest-warnings should be removed once we drop nose and we
+# rewrite tests using yield with parametrize
+addopts =
+    --doctest-modules
+    --disable-pytest-warnings
+
 [wheelhouse_uploader]
 artifact_indexes=
     # OSX wheels built by travis (only for specific tags):

diff --git a/sklearn/gaussian_process/tests/test_kernels.py b/sklearn/gaussian_process/tests/test_kernels.py
@@ -194,7 +194,6 @@ def check_hyperparameters_equal(kernel1, kernel2):
 
 def test_kernel_clone():
     # Test that sklearn's clone works correctly on kernels.
-    bounds = (1e-5, 1e5)
     for kernel in kernels:
         kernel_cloned = clone(kernel)
 
@@ -209,12 +208,17 @@ def test_kernel_clone():
         # Check that all hyperparameters are equal.
         yield check_hyperparameters_equal, kernel, kernel_cloned
 
-        # This test is to verify that using set_params does not
-        # break clone on kernels.
-        # This used to break because in kernels such as the RBF, non-trivial
-        # logic that modified the length scale used to be in the constructor
-        # See https://github.com/scikit-learn/scikit-learn/issues/6961
-        # for more details.
+
+def test_kernel_clone_after_set_params():
+    # This test is to verify that using set_params does not
+    # break clone on kernels.
+    # This used to break because in kernels such as the RBF, non-trivial
+    # logic that modified the length scale used to be in the constructor
+    # See https://github.com/scikit-learn/scikit-learn/issues/6961
+    # for more details.
+    bounds = (1e-5, 1e5)
+    for kernel in kernels:
+        kernel_cloned = clone(kernel)
         params = kernel.get_params()
         # RationalQuadratic kernel is isotropic.
         isotropic_kernels = (ExpSineSquared, RationalQuadratic)

diff --git a/sklearn/neighbors/tests/test_ball_tree.py b/sklearn/neighbors/tests/test_ball_tree.py
@@ -156,6 +156,14 @@ def compute_kernel_slow(Y, X, kernel, h):
         raise ValueError('kernel not recognized')
 
 
+def check_results(kernel, h, atol, rtol, breadth_first, bt, Y, dens_true):
+    dens = bt.kernel_density(Y, h, atol=atol, rtol=rtol,
+                             kernel=kernel,
+                             breadth_first=breadth_first)
+    assert_allclose(dens, dens_true,
+                    atol=atol, rtol=max(rtol, 1e-7))
+
+
 def test_ball_tree_kde(n_samples=100, n_features=3):
     np.random.seed(0)
     X = np.random.random((n_samples, n_features))
@@ -167,18 +175,11 @@ def test_ball_tree_kde(n_samples=100, n_features=3):
         for h in [0.01, 0.1, 1]:
             dens_true = compute_kernel_slow(Y, X, kernel, h)
 
-            def check_results(kernel, h, atol, rtol, breadth_first):
-                dens = bt.kernel_density(Y, h, atol=atol, rtol=rtol,
-                                         kernel=kernel,
-                                         breadth_first=breadth_first)
-                assert_allclose(dens, dens_true,
-                                atol=atol, rtol=max(rtol, 1e-7))
-
             for rtol in [0, 1E-5]:
                 for atol in [1E-6, 1E-2]:
                     for breadth_first in (True, False):
                         yield (check_results, kernel, h, atol, rtol,
-                               breadth_first)
+                               breadth_first, bt, Y, dens_true)
 
 
 def test_gaussian_kde(n_samples=1000):

diff --git a/sklearn/neighbors/tests/test_kd_tree.py b/sklearn/neighbors/tests/test_kd_tree.py
@@ -24,28 +24,29 @@ def brute_force_neighbors(X, Y, k, metric, **kwargs):
     return dist, ind
 
 
+def check_neighbors(dualtree, breadth_first, k, metric, X, Y, kwargs):
+    kdt = KDTree(X, leaf_size=1, metric=metric, **kwargs)
+    dist1, ind1 = kdt.query(Y, k, dualtree=dualtree,
+                            breadth_first=breadth_first)
+    dist2, ind2 = brute_force_neighbors(X, Y, k, metric, **kwargs)
+
+    # don't check indices here: if there are any duplicate distances,
+    # the indices may not match.  Distances should not have this problem.
+    assert_array_almost_equal(dist1, dist2)
+
+
 def test_kd_tree_query():
     np.random.seed(0)
     X = np.random.random((40, DIMENSION))
     Y = np.random.random((10, DIMENSION))
 
-    def check_neighbors(dualtree, breadth_first, k, metric, kwargs):
-        kdt = KDTree(X, leaf_size=1, metric=metric, **kwargs)
-        dist1, ind1 = kdt.query(Y, k, dualtree=dualtree,
-                                breadth_first=breadth_first)
-        dist2, ind2 = brute_force_neighbors(X, Y, k, metric, **kwargs)
-
-        # don't check indices here: if there are any duplicate distances,
-        # the indices may not match.  Distances should not have this problem.
-        assert_array_almost_equal(dist1, dist2)
-
     for (metric, kwargs) in METRICS.items():
         for k in (1, 3, 5):
             for dualtree in (True, False):
                 for breadth_first in (True, False):
                     yield (check_neighbors,
                            dualtree, breadth_first,
-                           k, metric, kwargs)
+                           k, metric, X, Y, kwargs)
 
 
 def test_kd_tree_query_radius(n_samples=100, n_features=10):
@@ -107,6 +108,14 @@ def compute_kernel_slow(Y, X, kernel, h):
         raise ValueError('kernel not recognized')
 
 
+def check_results(kernel, h, atol, rtol, breadth_first, Y, kdt, dens_true):
+    dens = kdt.kernel_density(Y, h, atol=atol, rtol=rtol,
+                              kernel=kernel,
+                              breadth_first=breadth_first)
+    assert_allclose(dens, dens_true, atol=atol,
+                    rtol=max(rtol, 1e-7))
+
+
 def test_kd_tree_kde(n_samples=100, n_features=3):
     np.random.seed(0)
     X = np.random.random((n_samples, n_features))
@@ -118,18 +127,11 @@ def test_kd_tree_kde(n_samples=100, n_features=3):
         for h in [0.01, 0.1, 1]:
             dens_true = compute_kernel_slow(Y, X, kernel, h)
 
-            def check_results(kernel, h, atol, rtol, breadth_first):
-                dens = kdt.kernel_density(Y, h, atol=atol, rtol=rtol,
-                                          kernel=kernel,
-                                          breadth_first=breadth_first)
-                assert_allclose(dens, dens_true, atol=atol,
-                                rtol=max(rtol, 1e-7))
-
             for rtol in [0, 1E-5]:
                 for atol in [1E-6, 1E-2]:
                     for breadth_first in (True, False):
                         yield (check_results, kernel, h, atol, rtol,
-                               breadth_first)
+                               breadth_first, Y, kdt, dens_true)
 
 
 def test_gaussian_kde(n_samples=1000):

diff --git a/sklearn/neighbors/tests/test_kde.py b/sklearn/neighbors/tests/test_kde.py
@@ -29,6 +29,17 @@ def compute_kernel_slow(Y, X, kernel, h):
         raise ValueError('kernel not recognized')
 
 
+def check_results(kernel, bandwidth, atol, rtol, X, Y, dens_true):
+    kde = KernelDensity(kernel=kernel, bandwidth=bandwidth,
+                        atol=atol, rtol=rtol)
+    log_dens = kde.fit(X).score_samples(Y)
+    assert_allclose(np.exp(log_dens), dens_true,
+                    atol=atol, rtol=max(1E-7, rtol))
+    assert_allclose(np.exp(kde.score(Y)),
+                    np.prod(dens_true),
+                    atol=atol, rtol=max(1E-7, rtol))
+
+
 def test_kernel_density(n_samples=100, n_features=3):
     rng = np.random.RandomState(0)
     X = rng.randn(n_samples, n_features)
@@ -39,20 +50,11 @@ def test_kernel_density(n_samples=100, n_features=3):
         for bandwidth in [0.01, 0.1, 1]:
             dens_true = compute_kernel_slow(Y, X, kernel, bandwidth)
 
-            def check_results(kernel, bandwidth, atol, rtol):
-                kde = KernelDensity(kernel=kernel, bandwidth=bandwidth,
-                                    atol=atol, rtol=rtol)
-                log_dens = kde.fit(X).score_samples(Y)
-                assert_allclose(np.exp(log_dens), dens_true,
-                                atol=atol, rtol=max(1E-7, rtol))
-                assert_allclose(np.exp(kde.score(Y)),
-                                np.prod(dens_true),
-                                atol=atol, rtol=max(1E-7, rtol))
-
             for rtol in [0, 1E-5]:
                 for atol in [1E-6, 1E-2]:
                     for breadth_first in (True, False):
-                        yield (check_results, kernel, bandwidth, atol, rtol)
+                        yield (check_results, kernel, bandwidth, atol, rtol,
+                               X, Y, dens_true)
 
 
 def test_kernel_density_sampling(n_samples=100, n_features=3):