[ENH] Venn Diagram: Add relations over columns, simplify over rows

Orange/widgets/visualize/tests/test_owvenndiagram.py

@@ -18,7 +18,8 @@
                                                     drop_columns,
                                                     OWVennDiagram,
                                                     group_table_indices,
-                                                    copy_descriptor)
+                                                    copy_descriptor,
+                                                    arrays_equal)
 from Orange.tests import test_filename
 
 
@@ -124,6 +125,8 @@ def _select_data(self):
         return list(range(len(self.signal_data)))
 
     def test_multiple_input(self):
+        """Over rows"""
+        self.widget.rowwise = True
         self.send_signal(self.signal_name, self.data[:100], 1)
         self.send_signal(self.signal_name, self.data[50:], 2)
 
@@ -161,6 +164,49 @@ def test_multiple_input(self):
         self.assertIsNone(self.get_output(self.widget.Outputs.selected_data))
         self.assertIsNone(self.get_output(self.widget.Outputs.annotated_data))
 
+    def test_multiple_input_over_cols(self):
+        self.widget.rowwise = False
+        selected_atr_name = 'Selected'
+        input2 = self.data.transform(Domain([self.data.domain.attributes[0]],
+                                            self.data.domain.class_vars,
+                                            self.data.domain.metas))
+        self.send_signal(self.signal_name, self.data, (1, 'Data', None))
+        self.send_signal(self.signal_name, input2, (2, 'Data', None))
+
+        selected = self.get_output(self.widget.Outputs.selected_data)
+        annotated = self.get_output(self.widget.Outputs.annotated_data)
+        self.assertIsNone(selected)
+        self.assertEqual(len(annotated), len(self.data))
+        self.assertEqual(annotated.domain, self.data.domain)
+        for atr in annotated.domain.attributes:
+            self.assertIsNotNone(atr.attributes[selected_atr_name])
+            self.assertFalse(atr.attributes[selected_atr_name])
+
+        # select data instances
+        self.widget.vennwidget.vennareas()[3].setSelected(True)
+        np.testing.assert_array_equal(self.get_output(self.widget.Outputs.selected_data).X,
+                                      input2.X)
+        np.testing.assert_array_equal(self.get_output(self.widget.Outputs.selected_data).Y,
+                                      input2.Y)
+        np.testing.assert_array_equal(self.get_output(self.widget.Outputs.selected_data).metas,
+                                      input2.metas)
+
+        #domain matches but the values do not
+        input2.X = input2.X - 1
+        self.send_signal(self.signal_name, input2, (2, 'Data', None))
+        self.widget.vennwidget.vennareas()[3].setSelected(True)
+        self.assertEqual(self.get_output(self.widget.Outputs.selected_data).domain.attributes[0].name,
+                         'sepal length-iris1')
+        self.assertEqual(self.get_output(self.widget.Outputs.selected_data).domain.attributes[1].name,
+                         'sepal length-iris2')
+
+        out_domain = self.get_output(self.widget.Outputs.annotated_data).domain.attributes
+        self.assertTrue(out_domain[0].attributes[selected_atr_name])
+        self.assertTrue(out_domain[1].attributes[selected_atr_name])
+        self.assertFalse(out_domain[2].attributes[selected_atr_name])
+        self.assertFalse(out_domain[3].attributes[selected_atr_name])
+        self.assertFalse(out_domain[4].attributes[selected_atr_name])
+
     def test_no_data(self):
         """Check that the widget doesn't crash on empty data"""
         self.send_signal(self.signal_name, self.data[:0], 1)
@@ -188,6 +234,51 @@ def test_unconditional_commit_on_new_signal(self):
             self.send_signal(self.signal_name, self.data[:100], 1)
             commit.assert_called()
 
+    def test_input_compatibility(self):
+        self.widget.rowwise = True
+        self.send_signal(self.signal_name, self.data, 1)
+        self.send_signal(self.signal_name,
+                         self.data.transform(Domain([self.data.domain.attributes[0]],
+                                                    self.data.domain.class_vars,
+                                                    self.data.domain.metas)), 2)
+        self.assertFalse(self.widget.Error.instances_mismatch.is_shown())
+
+        self.widget.rowwise = False
+        self.send_signal(self.signal_name, self.data[:100, :], 2)
+        self.assertTrue(self.widget.Error.instances_mismatch.is_shown())
+
+        self.send_signal(self.signal_name, None, 2)
+        self.assertFalse(self.widget.Error.instances_mismatch.is_shown())
+
+    def test_rows_identifiers(self):
+        self.widget.rowwise = True
+        data = Table('zoo')
+        self.send_signal(self.signal_name, data, (1, 'Data', None))
+        self.widget.selected_feature = 'name'
+        self.send_signal(self.signal_name, data[:5], (2, 'Data', None))
+
+        self.assertIsNone(self.get_output(self.widget.Outputs.selected_data))
+        self.widget.vennwidget.vennareas()[3].setSelected(True)
+        selected = self.get_output(self.widget.Outputs.selected_data)
+        self.assertEqual(len(selected), 5)
+        self.assertEqual(selected.domain.attributes, data.domain.attributes)
+        self.assertEqual(selected.domain.class_vars, data.domain.class_vars)
+
+        annotated = self.get_output(self.widget.Outputs.annotated_data)
+        self.assertEqual(len(annotated), 100)
+
+    def test_too_many_inputs(self):
+        self.send_signal(self.signal_name, self.data, 1)
+        self.send_signal(self.signal_name, self.data, 2)
+        self.send_signal(self.signal_name, self.data, 3)
+        self.send_signal(self.signal_name, self.data, 4)
+        self.send_signal(self.signal_name, self.data, 5)
+        self.send_signal(self.signal_name, self.data, 6)
+        self.assertTrue(self.widget.Error.too_many_inputs.is_shown())
+
+        self.send_signal(self.signal_name, None, 6)
+        self.assertFalse(self.widget.Error.too_many_inputs.is_shown())
+
 
 class GroupTableIndicesTest(unittest.TestCase):
 
@@ -278,6 +369,21 @@ class SomeVariable(Variable):
         copied = copy_descriptor(var, "cux")
         self.assertEqual(copied.name, "cux")
 
+    def test_array_equals(self):
+        a = np.array([1, 2], dtype=np.float64)
+        b = np.array([1, np.nan], dtype=np.float64)
+        self.assertTrue(arrays_equal(None, None, None))
+        self.assertFalse(arrays_equal(None, a, None))
+        self.assertFalse(arrays_equal(a, None, None))
+        self.assertFalse(arrays_equal(a, b, ContinuousVariable))
+        a[1] = np.nan
+        self.assertTrue(arrays_equal(a, b, ContinuousVariable))
+        self.assertTrue(arrays_equal(a.astype(str), a.astype(str), StringVariable))
+        a[1] = 2
+        b[1] = 3
+        self.assertFalse(arrays_equal(a, b, ContinuousVariable))
+        self.assertFalse(arrays_equal(a.astype(str), b.astype(str), StringVariable))
+
 
 if __name__ == "__main__":
     unittest.main()

doc/visual-programming/source/widgets/visualize/venndiagram.md

@@ -10,19 +10,18 @@ Plots a [Venn diagram](http://en.wikipedia.org/wiki/Venn_diagram) for two or mor
 **Outputs**
 
 - Selected Data: instances selected from the plot
+- Data: entire data with a column indicating whether an instance was selected or not
 
-The **Venn Diagram** widget displays logical relations between datasets. This projection shows two or more datasets represented by circles of different colors. The intersections are subsets that belong to more than one dataset. To further analyze or visualize the subset, click on the intersection.
+The **Venn Diagram** widget displays logical relations between datasets by showing the number of common data instances (rows) or the number of shared features (columns). Selecting a part of the visualization outputs the corresponding instances or features.
 
 ![](images/venn-workflow.png)
 
-![](images/venn-identifiers-stamped.png)
+![](images/VennDiagram-stamped.png)
 
-1. Information on the input data.
-2. Select the identifiers by which to compare the data.
-3. Tick *Output duplicates* if you wish to remove duplicates.
-4. If *Auto commit* is on, changes are automatically communicated to other widgets. Alternatively, click *Commit*.
-5. *Save Image* saves the created image to your computer in a .svg or .png format.
-6. Produce a report.
+1. Select whether to count common features or instances.
+2. Select whether to include duplicates or to output only unique rows (applicable only when matching by instances). If *Auto commit* is on, changes are automatically communicated to other widgets.
+
+Rows can be matched by their identity, e.g. rows from different data sets match if they came from the same row in a file. Instead of using identities, we can choose a string variable to match the rows by. A warning is shown if data sets have no common string variable.
 
 Examples
 --------