Merge pull request #3737 from pavlin-policar/kmeans-louvain-array-eq

Replace statistics.util.array_equal with data.util.array_equal

Author: janezd

Orange/statistics/util.py

@@ -7,28 +7,18 @@
 import warnings
 from typing import Iterable
 
-import numpy as np
 import bottleneck as bn
-from scipy import sparse as sp
+import numpy as np
 import scipy.stats.stats
-from sklearn.utils.sparsefuncs import mean_variance_axis
-
-
-def sparse_array_equal(x1, x2):
-    """Check if two sparse arrays are equal."""
-    if not sp.issparse(x1):
-        raise TypeError("`x1` must be sparse.")
-    if not sp.issparse(x2):
-        raise TypeError("`x2` must be sparse.")
+from scipy import sparse as sp
 
-    return x1.shape == x2.shape and (x1 != x2).nnz == 0
+from sklearn.utils.sparsefuncs import mean_variance_axis
 
+from Orange.data.util import array_equal
+from Orange.util import deprecated
 
-def array_equal(x1, x2):
-    """Equivalent of np.array_equal that properly handles sparse matrices."""
-    if sp.issparse(x1) and sp.issparse(x2):
-        return sparse_array_equal(x1, x2)
-    return np.array_equal(x1, x2)
+# For backwards compatibility
+array_equal = deprecated("Orange.data.util.array_equal")(array_equal)
 
 
 def _count_nans_per_row_sparse(X, weights, dtype=None):

Orange/tests/test_statistics.py

@@ -1,15 +1,17 @@
 import unittest
 import warnings
+from distutils.version import LooseVersion
 from itertools import chain
 from functools import partial, wraps
 
 import numpy as np
 from scipy.sparse import csr_matrix, issparse, lil_matrix, csc_matrix, \
     SparseEfficiencyWarning
 
+import Orange
 from Orange.statistics.util import bincount, countnans, contingency, digitize, \
     mean, nanmax, nanmean, nanmedian, nanmin, nansum, nanunique, stats, std, \
-    unique, var, nanstd, nanvar, nanmode, array_equal, FDR
+    unique, var, nanstd, nanvar, nanmode, FDR
 from sklearn.utils import check_random_state
 
 
@@ -34,10 +36,20 @@ def sparse_with_explicit_zero(x, array):
 
             return sp_array
 
+        # Make sure to call setUp and tearDown methods in between test runs so
+        # any widget state doesn't interfere between tests
+        def _setup_teardown():
+            self.tearDown()
+            self.setUp()
+
         test_case(self, np.array)
+        _setup_teardown()
         test_case(self, csr_matrix)
+        _setup_teardown()
         test_case(self, csc_matrix)
+        _setup_teardown()
         test_case(self, partial(sparse_with_explicit_zero, array=csr_matrix))
+        _setup_teardown()
         test_case(self, partial(sparse_with_explicit_zero, array=csc_matrix))
 
     return _wrapper
@@ -295,6 +307,17 @@ def test_FDR_list(self):
             np.array([0.00033, 0.0004, 0.00005, 0.00038, 0.00025]),
             result, decimal=5)
 
+    def test_array_equal_deprecated(self):
+        """This test is to be included in the 3.22 release and will fail in
+        version 3.24. This serves as a reminder to remove the deprecated method
+        and this test."""
+        if LooseVersion(Orange.__version__) >= LooseVersion("3.24"):
+            self.fail(
+                "`Orange.statistics.util.array_equal` was deprecated in "
+                "version 3.22, and there have been two minor versions in "
+                "between. Please remove the deprecated method from the module."
+            )
+
 
 class TestNanmean(unittest.TestCase):
     def setUp(self):
@@ -622,25 +645,6 @@ def test_nanunique_ignores_nans_in_counts(self, array):
         np.testing.assert_equal(nanunique(x, return_counts=True)[1], expected)
 
 
-class TestArrayEqual(unittest.TestCase):
-    @dense_sparse
-    def test_same_matrices(self, array):
-        x = array([0, 1, 0, 0, 2])
-        self.assertTrue(array_equal(x, x))
-
-    @dense_sparse
-    def test_with_different_shapes(self, array):
-        x = array(np.eye(4))
-        y = array(np.eye(5))
-        self.assertFalse(array_equal(x, y))
-
-    @dense_sparse
-    def test_with_different_values(self, array):
-        x = array([0, 1, 0, 0, 2])
-        y = array([0, 3, 0, 0, 2])
-        self.assertFalse(array_equal(x, y))
-
-
 class TestNanModeAppVeyor(unittest.TestCase):
     def test_appveyour_still_not_onscipy_1_2_0(self):
         import scipy

Orange/widgets/tests/utils.py

@@ -336,7 +336,11 @@ def table_dense_sparse(test_case):
 
     @wraps(test_case)
     def _wrapper(self):
+        # Make sure to call setUp and tearDown methods in between test runs so
+        # any widget state doesn't interfere between tests
         test_case(self, lambda table: table.to_dense())
+        self.tearDown()
+        self.setUp()
         test_case(self, lambda table: table.to_sparse())
 
     return _wrapper

Orange/widgets/unsupervised/owkmeans.py

@@ -10,7 +10,7 @@
 from Orange.clustering import KMeans
 from Orange.clustering.kmeans import KMeansModel, SILHOUETTE_MAX_SAMPLES
 from Orange.data import Table, Domain, DiscreteVariable, ContinuousVariable
-from Orange.data.util import get_unique_names
+from Orange.data.util import get_unique_names, array_equal
 from Orange.preprocess.impute import ReplaceUnknowns
 from Orange.widgets import widget, gui
 from Orange.widgets.settings import Setting
@@ -526,7 +526,7 @@ def set_data(self, data):
         self.data, old_data = data, self.data
 
         # Do not needlessly recluster the data if X hasn't changed
-        if old_data and self.data and np.array_equal(self.data.X, old_data.X):
+        if old_data and self.data and array_equal(self.data.X, old_data.X):
             if self.auto_commit:
                 self.send_data()
         else:

Orange/widgets/unsupervised/owlouvainclustering.py

@@ -16,10 +16,9 @@
 
 from Orange.clustering.louvain import table_to_knn_graph, Louvain
 from Orange.data import Table, DiscreteVariable
-from Orange.data.util import get_unique_names
+from Orange.data.util import get_unique_names, array_equal
 from Orange import preprocess
 from Orange.projection import PCA
-from Orange.statistics import util as ut
 from Orange.widgets import widget, gui, report
 from Orange.widgets.settings import DomainContextHandler, ContextSetting, \
     Setting
@@ -408,7 +407,7 @@ def set_data(self, data):
         # Make sure to properly enable/disable slider based on `apply_pca` setting
         self.controls.pca_components.setEnabled(self.apply_pca)
 
-        if prev_data and self.data and ut.array_equal(prev_data.X, self.data.X):
+        if prev_data and self.data and array_equal(prev_data.X, self.data.X):
             if self.auto_commit:
                 self._send_data()
             return

Orange/widgets/unsupervised/owmds.py

@@ -9,6 +9,7 @@
 import pyqtgraph as pg
 
 from Orange.data import ContinuousVariable, Domain, Table, StringVariable
+from Orange.data.util import array_equal
 from Orange.distance import Euclidean
 from Orange.misc import DistMatrix
 from Orange.projection.manifold import torgerson, MDS
@@ -323,8 +324,7 @@ def _initialize(self):
         self.openContext(self.data)
         self._invalidated = not (matrix_existed and
                                  self.effective_matrix is not None and
-                                 np.array_equal(effective_matrix,
-                                                self.effective_matrix))
+                                 array_equal(effective_matrix, self.effective_matrix))
         if self._invalidated:
             self.clear()
         self.graph.set_effective_matrix(self.effective_matrix)