Merge pull request #738 from markotoplak/refactor-preprocess-tests

[MNT] Refactor preprocess tests

orangecontrib/spectroscopy/tests/test_als.py

@@ -4,9 +4,19 @@
 
 from Orange.data import Table
 from orangecontrib.spectroscopy.preprocess.als import ALSP, ARPLS, AIRPLS
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALLER_COLLAGEN
 
 
-class Testals(unittest.TestCase):
+class TestAls(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = [
+        ALSP(lam=100E+6, itermax=5, p=0.5),
+        ARPLS(lam=100E+5, itermax=5, ratio=0.5),
+        AIRPLS(lam=100, itermax=5, porder=1),
+    ]
+    data = SMALLER_COLLAGEN
+
     def test_als_Basic(self):
         data = Table.from_numpy(None, [[1.0, 2.0, 10.0, 5.0],
                                        [3.0, 5.0, 9.0, 4.0]])

orangecontrib/spectroscopy/tests/test_atm_corr.py

@@ -4,9 +4,16 @@
 
 from orangecontrib.spectroscopy.preprocess.atm_corr import AtmCorr
 from orangecontrib.spectroscopy.tests.util import spectra_table
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALL_COLLAGEN, add_edge_case_data_parameter
 
+class TestAtmCorr(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = list(
+        add_edge_case_data_parameter(AtmCorr, "reference", SMALL_COLLAGEN[0:1],
+                                     correct_ranges=[(1300, 2100)], smooth_win=5))
+    data = SMALL_COLLAGEN
 
-class TestAtmCorr(unittest.TestCase):
     def test_atm_corr(self):
         # Fake atmospheric spectrum
         def atm(wn):

orangecontrib/spectroscopy/tests/test_conversion.py

@@ -10,19 +10,17 @@
 from Orange.evaluation.scoring import AUC
 from Orange.data.table import DomainTransformationError
 
-from orangecontrib.spectroscopy.tests.test_preprocess import \
-    PREPROCESSORS_INDEPENDENT_SAMPLES, \
-    PREPROCESSORS
-
-from orangecontrib.spectroscopy.tests.test_preprocess import SMALL_COLLAGEN, preprocessor_data
-
 from orangecontrib.spectroscopy.preprocess import Interpolate, \
     Cut, SavitzkyGolayFiltering
 from orangecontrib.spectroscopy.data import getx
+from orangecontrib.spectroscopy.tests.util import smaller_data
 
 
 logreg = LogisticRegressionLearner(max_iter=1000)
 
+COLLAGEN = Orange.data.Table("collagen")
+SMALL_COLLAGEN = smaller_data(COLLAGEN, 2, 2)
+
 
 def separate_learn_test(data):
     sf = ms.ShuffleSplit(n_splits=1, test_size=0.2, random_state=np.random.RandomState(0))
@@ -80,20 +78,6 @@ def test_predict_different_domain_interpolation(self):
         # the more we cut the lower precision we get
         self.assertTrue(aucorig > auccut1 > auccut2 > auccut3)
 
-    def test_whole_and_train_separete(self):
-        """ Applying a preprocessor before spliting data into train and test
-        and applying is just on train data should yield the same transformation of
-        the test data. """
-        for proc in PREPROCESSORS_INDEPENDENT_SAMPLES:
-            with self.subTest(proc):
-                data = preprocessor_data(proc)
-                _, test1 = separate_learn_test(proc(data))
-                train, test = separate_learn_test(data)
-                train = proc(train)
-                test_transformed = test.transform(train.domain)
-                np.testing.assert_almost_equal(test_transformed.X, test1.X,
-                                               err_msg="Preprocessor " + str(proc))
-
     def test_predict_savgov_same_domain(self):
         data = SavitzkyGolayFiltering(window=9, polyorder=2, deriv=2)(self.collagen)
         train, test = separate_learn_test(data)
@@ -107,44 +91,3 @@ def test_predict_savgol_another_interpolate(self):
         train = Interpolate(points=getx(train))(train)
         aucai = AUC(TestOnTestData()(train, test, [logreg]))
         self.assertAlmostEqual(auc, aucai, delta=0.02)
-
-    def test_slightly_different_domain(self):
-        """ If test data has a slightly different domain then (with interpolation)
-        we should obtain a similar classification score. """
-        # rows full of unknowns make LogisticRegression undefined
-        # we can obtain them, for example, with EMSC, if one of the badspectra
-        # is a spectrum from the data
-        learner = LogisticRegressionLearner(max_iter=1000, preprocessors=[_RemoveNaNRows()])
-
-        for proc in PREPROCESSORS:
-            if hasattr(proc, "skip_add_zeros"):
-                continue
-            with self.subTest(proc):
-                # LR that can not handle unknown values
-                train, test = separate_learn_test(preprocessor_data(proc))
-                train1 = proc(train)
-                aucorig = AUC(TestOnTestData()(train1, test, [learner]))
-                test = slightly_change_wavenumbers(test, 0.00001)
-                test = odd_attr(test)
-                # a subset of points for training so that all test sets points
-                # are within the train set points, which gives no unknowns
-                train = Interpolate(points=getx(train)[1:-3])(train)  # interpolatable train
-                train = proc(train)
-                # explicit domain conversion test to catch exceptions that would
-                # otherwise be silently handled in TestOnTestData
-                _ = test.transform(train.domain)
-                aucnow = AUC(TestOnTestData()(train, test, [learner]))
-                self.assertAlmostEqual(aucnow, aucorig, delta=0.03, msg="Preprocessor " + str(proc))
-                test = Interpolate(points=getx(test) - 1.)(test)  # also do a shift
-                _ = test.transform(train.domain)  # explicit call again
-                aucnow = AUC(TestOnTestData()(train, test, [learner]))
-                # the difference should be slight
-                self.assertAlmostEqual(aucnow, aucorig, delta=0.05, msg="Preprocessor " + str(proc))
-
-
-class _RemoveNaNRows(Orange.preprocess.preprocess.Preprocess):
-
-    def __call__(self, data):
-        mask = np.isnan(data.X)
-        mask = np.any(mask, axis=1)
-        return data[~mask]

orangecontrib/spectroscopy/tests/test_cut.py

@@ -3,9 +3,14 @@
 import unittest
 from orangecontrib.spectroscopy.preprocess import Cut
 from orangecontrib.spectroscopy.data import getx
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALL_COLLAGEN
 
 
-class TestCut(unittest.TestCase):
+class TestCut(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = [Cut(lowlim=1000, highlim=1800)]
+    data = SMALL_COLLAGEN
 
     @classmethod
     def setUpClass(cls):

orangecontrib/spectroscopy/tests/test_despike.py

@@ -2,9 +2,15 @@
 import numpy as np
 from Orange.data import Table
 from orangecontrib.spectroscopy.preprocess import Despike
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALL_COLLAGEN
 
 
-class TestSpikeremoval(unittest.TestCase):
+class TestSpikeRemoval(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = [Despike(threshold=5, cutoff=60, dis=5)]
+    data = SMALL_COLLAGEN
+
     def test_spikes(self):
         data = Table.from_numpy(None, [[1000, 1, 1, 1, 1, 10, 1, 1, 1000, 1000, 1000, 1, 1000,
                                         1, 1, 1, 1000, 1000, 1000, 1000],

orangecontrib/spectroscopy/tests/test_emsc.py

@@ -8,9 +8,19 @@
     SelectionFunction, SmoothedSelectionFunction
 from orangecontrib.spectroscopy.preprocess.npfunc import Sum
 from orangecontrib.spectroscopy.tests.util import spectra_table
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALL_COLLAGEN, add_edge_case_data_parameter
 
 
-class TestEMSC(unittest.TestCase):
+class TestEMSC(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    different_reference = list(
+        add_edge_case_data_parameter(EMSC, "reference", SMALL_COLLAGEN[0:1]))
+    different_badspectra = list(
+        add_edge_case_data_parameter(EMSC, "badspectra", SMALL_COLLAGEN[0:2],
+                                     reference=SMALL_COLLAGEN[-1:]))
+    preprocessors = different_reference + different_badspectra
+    data = SMALL_COLLAGEN
 
     def test_ab(self):
         data = Table.from_numpy(None, [[1.0, 2.0, 1.0, 1.0],

orangecontrib/spectroscopy/tests/test_integrate.py

@@ -4,9 +4,24 @@
 import numpy as np
 
 from orangecontrib.spectroscopy.preprocess import Integrate
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALL_COLLAGEN
 
 
-class TestIntegrate(unittest.TestCase):
+class TestIntegrate(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = [
+        Integrate(limits=[[900, 100], [1100, 1200], [1200, 1300]]),
+        Integrate(methods=Integrate.Simple, limits=[[1100, 1200]]),
+        Integrate(methods=Integrate.Baseline, limits=[[1100, 1200]]),
+        Integrate(methods=Integrate.PeakMax, limits=[[1100, 1200]]),
+        Integrate(methods=Integrate.PeakBaseline, limits=[[1100, 1200]]),
+        Integrate(methods=Integrate.PeakAt, limits=[[1100]]),
+        Integrate(methods=Integrate.PeakX, limits=[[1100, 1200]]),
+        Integrate(methods=Integrate.PeakXBaseline, limits=[[1100, 1200]])
+    ]
+    data = SMALL_COLLAGEN
+
 
     def test_simple(self):
         data = Table.from_numpy(None, [[1, 2, 3, 1, 1, 1],

orangecontrib/spectroscopy/tests/test_interpolate.py

@@ -16,9 +16,14 @@
     nan_extend_edges_and_interpolate
 from orangecontrib.spectroscopy.data import getx
 from orangecontrib.spectroscopy.tests.util import spectra_table
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALL_COLLAGEN
 
 
-class TestInterpolate(unittest.TestCase):
+class TestInterpolate(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = [Interpolate(np.linspace(1000, 1700, 100))]
+    data = SMALL_COLLAGEN
 
     @classmethod
     def setUpClass(cls):

orangecontrib/spectroscopy/tests/test_me_emsc.py

@@ -8,6 +8,8 @@
 from orangecontrib.spectroscopy.preprocess.me_emsc import ME_EMSC
 from orangecontrib.spectroscopy.preprocess.emsc import SelectionFunction, SmoothedSelectionFunction
 from orangecontrib.spectroscopy.preprocess.npfunc import Sum
+from orangecontrib.spectroscopy.tests.test_preprocess import TestCommonIndpSamplesMixin, \
+    SMALLER_COLLAGEN, add_edge_case_data_parameter
 
 
 def weights_from_inflection_points_legacy(points, kappa, wavenumbers):
@@ -82,7 +84,11 @@ def weights_from_inflection_points_legacy(points, kappa, wavenumbers):
     return data
 
 
-class TestME_EMSC(unittest.TestCase):
+class TestME_EMSC(unittest.TestCase, TestCommonIndpSamplesMixin):
+
+    preprocessors = list(
+        add_edge_case_data_parameter(ME_EMSC, "reference", SMALLER_COLLAGEN[0:1], max_iter=4))
+    data = SMALLER_COLLAGEN
 
     @classmethod
     def setUpClass(cls):