Gradient Boosting - Remove try/catch around imports

Orange/classification/__init__.py

@@ -20,12 +20,6 @@
 from .sgd import *
 from .neural_network import *
 from .calibration import *
-try:
-    from .catgb import *
-except ModuleNotFoundError:
-    pass
+from .catgb import *
 from .gb import *
-try:
-    from .xgb import *
-except Exception:
-    pass
+from .xgb import *

Orange/modelling/__init__.py

@@ -11,12 +11,6 @@
 from .randomforest import *
 from .svm import *
 from .tree import *
-try:
-    from .catgb import *
-except ImportError:
-    pass
+from .catgb import *
 from .gb import *
-try:
-    from .xgb import *
-except ImportError:
-    pass
+from .xgb import *

Orange/regression/__init__.py

@@ -14,13 +14,7 @@
 from .tree import *
 from .neural_network import *
 from ..classification.simple_tree import *
-try:
-    from .catgb import *
-except ModuleNotFoundError:
-    pass
+from .catgb import *
 from .gb import *
-try:
-    from .xgb import *
-except Exception:
-    pass
+from .xgb import *
 from .curvefit import *

Orange/widgets/model/owgradientboosting.py

@@ -9,44 +9,26 @@
 from Orange.base import Learner
 from Orange.data import Table
 from Orange.modelling import GBLearner
-
-try:
-    from Orange.modelling import CatGBLearner
-except ImportError:
-    CatGBLearner = None
-try:
-    from Orange.modelling import XGBLearner, XGBRFLearner
-except ImportError:
-    XGBLearner = XGBRFLearner = None
-
+from Orange.modelling import CatGBLearner
+from Orange.modelling import XGBLearner, XGBRFLearner
 from Orange.widgets import gui
 from Orange.widgets.settings import Setting, SettingProvider
 from Orange.widgets.utils.owlearnerwidget import OWBaseLearner
 from Orange.widgets.utils.widgetpreview import WidgetPreview
 
 
 class LearnerItemModel(QStandardItemModel):
-    LEARNERS = [
-        (GBLearner, "", ""),
-        (XGBLearner, "Extreme Gradient Boosting (xgboost)", "xgboost"),
-        (XGBRFLearner, "Extreme Gradient Boosting Random Forest (xgboost)",
-         "xgboost"),
-        (CatGBLearner, "Gradient Boosting (catboost)", "catboost"),
-    ]
+    LEARNERS = [GBLearner, XGBLearner, XGBRFLearner, CatGBLearner]
 
     def __init__(self, parent):
         super().__init__(parent)
         self._add_data()
 
     def _add_data(self):
-        for cls, opt_name, lib in self.LEARNERS:
+        for cls in self.LEARNERS:
             item = QStandardItem()
-            imported = bool(cls)
-            name = cls.name if imported else opt_name
+            name = cls.name
             item.setData(f"{name}", Qt.DisplayRole)
-            item.setEnabled(imported)
-            if not imported:
-                item.setToolTip(f"{lib} is not installed")
             self.appendRow(item)
 
 
@@ -143,7 +125,7 @@ def _add_main_layout(self):
         self._set_lambda_label()
 
     def _set_lambda_label(self):
-        self.lambda_label.setText("Lambda: {}".format(self.lambda_))
+        self.lambda_label.setText(f"Lambda: {self.lambda_}")
 
     def get_arguments(self) -> Dict:
         params = super().get_arguments()

Orange/widgets/model/tests/test_owgradientboosting.py

@@ -1,32 +1,18 @@
 import json
 import unittest
-import sys
 from typing import Type
-from unittest.mock import patch, Mock
+from unittest.mock import Mock
 
 from Orange.classification import GBClassifier
 
-try:
-    from Orange.classification import XGBClassifier, XGBRFClassifier
-except ImportError:
-    XGBClassifier = XGBRFClassifier = None
-try:
-    from Orange.classification import CatGBClassifier
-except ImportError:
-    CatGBClassifier = None
+from Orange.classification import XGBClassifier, XGBRFClassifier
+from Orange.classification import CatGBClassifier
 from Orange.data import Table
-from Orange.modelling import GBLearner
 from Orange.preprocess.score import Scorer
 from Orange.regression import GBRegressor
 
-try:
-    from Orange.regression import XGBRegressor, XGBRFRegressor
-except ImportError:
-    XGBRegressor = XGBRFRegressor = None
-try:
-    from Orange.regression import CatGBRegressor
-except ImportError:
-    CatGBRegressor = None
+from Orange.regression import XGBRegressor, XGBRFRegressor
+from Orange.regression import CatGBRegressor
 from Orange.widgets.model.owgradientboosting import OWGradientBoosting, \
     LearnerItemModel, GBLearnerEditor, XGBLearnerEditor, XGBRFLearnerEditor, \
     CatGBLearnerEditor, BaseEditor
@@ -65,16 +51,6 @@ def test_model(self):
             self.assertEqual(model.item(i).isEnabled(),
                              classifiers[i] is not None)
 
-    @patch("Orange.widgets.model.owgradientboosting.LearnerItemModel.LEARNERS",
-           [(GBLearner, "", ""),
-            (None, "Gradient Boosting (catboost)", "catboost")])
-    def test_missing_lib(self):
-        widget = create_parent(CatGBLearnerEditor)
-        model = LearnerItemModel(widget)
-        self.assertEqual(model.rowCount(), 2)
-        self.assertTrue(model.item(0).isEnabled())
-        self.assertFalse(model.item(1).isEnabled())
-
 
 class BaseEditorTest(GuiTest):
     EditorClass: Type[BaseEditor] = None
@@ -146,7 +122,6 @@ def test_arguments(self):
                 "colsample_bynode": 1, "subsample": 1, "random_state": 0}
         self.assertDictEqual(self.editor.get_arguments(), args)
 
-    @unittest.skipIf(XGBClassifier is None, "Missing 'xgboost' package")
     def test_learner_parameters(self):
         params = (("Method", "Extreme Gradient Boosting (xgboost)"),
                   ("Number of trees", 100),
@@ -160,7 +135,6 @@ def test_learner_parameters(self):
                   ("Fraction of features for each split", 1))
         self.assertTupleEqual(self.editor.get_learner_parameters(), params)
 
-    @unittest.skipIf(XGBClassifier is None, "Missing 'xgboost' package")
     def test_default_parameters_cls(self):
         data = Table("heart_disease")
         booster = XGBClassifier()
@@ -178,7 +152,6 @@ def test_default_parameters_cls(self):
         self.assertEqual(int(tp["colsample_bylevel"]), self.editor.colsample_bylevel)
         self.assertEqual(int(tp["colsample_bynode"]), self.editor.colsample_bynode)
 
-    @unittest.skipIf(XGBRegressor is None, "Missing 'xgboost' package")
     def test_default_parameters_reg(self):
         data = Table("housing")
         booster = XGBRegressor()
@@ -206,7 +179,6 @@ def test_arguments(self):
                 "colsample_bynode": 1, "subsample": 1, "random_state": 0}
         self.assertDictEqual(self.editor.get_arguments(), args)
 
-    @unittest.skipIf(XGBRFClassifier is None, "Missing 'xgboost' package")
     def test_learner_parameters(self):
         params = (("Method",
                    "Extreme Gradient Boosting Random Forest (xgboost)"),
@@ -221,7 +193,6 @@ def test_learner_parameters(self):
                   ("Fraction of features for each split", 1))
         self.assertTupleEqual(self.editor.get_learner_parameters(), params)
 
-    @unittest.skipIf(XGBRFClassifier is None, "Missing 'xgboost' package")
     def test_default_parameters_cls(self):
         data = Table("heart_disease")
         booster = XGBRFClassifier()
@@ -239,7 +210,6 @@ def test_default_parameters_cls(self):
         self.assertEqual(int(tp["colsample_bylevel"]), self.editor.colsample_bylevel)
         self.assertEqual(int(tp["colsample_bynode"]), self.editor.colsample_bynode)
 
-    @unittest.skipIf(XGBRFRegressor is None, "Missing 'xgboost' package")
     def test_default_parameters_reg(self):
         data = Table("housing")
         booster = XGBRFRegressor()
@@ -266,7 +236,6 @@ def test_arguments(self):
                 "reg_lambda": 3, "colsample_bylevel": 1, "random_state": 0}
         self.assertDictEqual(self.editor.get_arguments(), args)
 
-    @unittest.skipIf(CatGBClassifier is None, "Missing 'catboost' package")
     def test_learner_parameters(self):
         params = (("Method", "Gradient Boosting (catboost)"),
                   ("Number of trees", 100),
@@ -277,7 +246,6 @@ def test_learner_parameters(self):
                   ("Fraction of features for each tree", 1))
         self.assertTupleEqual(self.editor.get_learner_parameters(), params)
 
-    @unittest.skipIf(CatGBClassifier is None, "Missing 'catboost' package")
     def test_default_parameters_cls(self):
         data = Table("heart_disease")
         booster = CatGBClassifier()
@@ -291,7 +259,6 @@ def test_default_parameters_cls(self):
         self.assertEqual(self.editor.learning_rate, 0.3)
         # params["learning_rate"] is automatically defined so don't test it
 
-    @unittest.skipIf(CatGBRegressor is None, "Missing 'catboost' package")
     def test_default_parameters_reg(self):
         data = Table("housing")
         booster = CatGBRegressor()
@@ -305,6 +272,7 @@ def test_default_parameters_reg(self):
         self.assertEqual(self.editor.learning_rate, 0.3)
         # params["learning_rate"] is automatically defined so don't test it
 
+
 class TestOWGradientBoosting(WidgetTest, WidgetLearnerTestMixin):
     def setUp(self):
         self.widget = self.create_widget(OWGradientBoosting,
@@ -328,7 +296,6 @@ def test_datasets(self):
         for ds in datasets.datasets():
             self.send_signal(self.widget.Inputs.data, ds)
 
-    @unittest.skipIf(XGBClassifier is None, "Missing 'xgboost' package")
     def test_xgb_params(self):
         simulate.combobox_activate_index(self.widget.controls.method_index, 1)
         editor = self.widget.editor
@@ -350,27 +317,11 @@ def test_xgb_params(self):
     def test_methods(self):
         self.send_signal(self.widget.Inputs.data, self.data)
         method_cb = self.widget.controls.method_index
-        for i, (cls, _, _) in enumerate(LearnerItemModel.LEARNERS):
-            if cls is None:
-                continue
+        for i, cls in enumerate(LearnerItemModel.LEARNERS):
             simulate.combobox_activate_index(method_cb, i)
             self.click_apply()
             self.assertIsInstance(self.widget.learner, cls)
 
-    def test_missing_lib(self):
-        modules = {k: v for k, v in sys.modules.items()
-                   if "orange" not in k.lower()}  # retain built-ins
-        modules["xgboost"] = None
-        modules["catboost"] = None
-        # pylint: disable=reimported,redefined-outer-name
-        # pylint: disable=import-outside-toplevel
-        with patch.dict(sys.modules, modules, clear=True):
-            from Orange.widgets.model.owgradientboosting import \
-                OWGradientBoosting
-            widget = self.create_widget(OWGradientBoosting,
-                                        stored_settings={"method_index": 3})
-            self.assertEqual(widget.method_index, 0)
-
 
 if __name__ == "__main__":
     unittest.main()

i18n/si.jaml

@@ -9596,14 +9596,8 @@ widgets/model/owcurvefit.py:
     housing: false
 widgets/model/owgradientboosting.py:
     class `LearnerItemModel`:
-        Extreme Gradient Boosting (xgboost): true
-        xgboost: false
-        Extreme Gradient Boosting Random Forest (xgboost): true
-        Gradient Boosting (catboost): true
-        catboost: false
         def `_add_data`:
             {name}: false
-            {lib} is not installed: {lib} ni nameščen
     class `BaseEditor`:
         def `_add_main_layout`:
             callback: false
@@ -9638,7 +9632,7 @@ widgets/model/owgradientboosting.py:
             Regularization:: Regularizacija
             lambda_index: false
         def `_set_lambda_label`:
-            'Lambda: {}': true
+            'Lambda: {self.lambda_}': true
         def `get_arguments`:
             reg_lambda: false
         def `get_learner_parameters`:
@@ -14893,6 +14887,8 @@ widgets/visualize/utils/customizableplot.py:
             {font.family()}: false
             font-style: false
             {fstyle}: false
+            foreground: false
+            color: false
         def `update_axes_ticks_font`:
             tickFont: false
         def `update_legend_font`: