Merge pull request #3346 from VesnaT/owtransform

[ENH] Transform: Add new widget

Author: BlazZupan

Orange/preprocess/preprocess.py

@@ -488,6 +488,18 @@ def transform(var):
         return data.transform(domain)
 
 
+class ApplyDomain(Preprocess):
+    def __init__(self, domain, name):
+        self._domain = domain
+        self._name = name
+
+    def __call__(self, data):
+        return data.transform(self._domain)
+
+    def __str__(self):
+        return self._name
+
+
 class PreprocessorList(Preprocess):
     """
     Store a list of preprocessors and on call apply them to the dataset.

Orange/widgets/data/icons/Transform.svg

@@ -0,0 +1,112 @@
+from Orange.data import Table
+from Orange.preprocess.preprocess import Preprocess, Discretize
+from Orange.widgets import gui
+from Orange.widgets.utils.sql import check_sql_input
+from Orange.widgets.widget import OWWidget, Input, Output, Msg
+
+
+class OWTransform(OWWidget):
+    name = "Transform"
+    description = "Transform data table."
+    icon = "icons/Transform.svg"
+    priority = 2110
+    keywords = []
+
+    class Inputs:
+        data = Input("Data", Table, default=True)
+        preprocessor = Input("Preprocessor", Preprocess)
+
+    class Outputs:
+        transformed_data = Output("Transformed Data", Table)
+
+    class Error(OWWidget.Error):
+        pp_error = Msg("An error occurred while transforming data.\n{}")
+
+    resizing_enabled = False
+    want_main_area = False
+
+    def __init__(self):
+        super().__init__()
+        self.data = None
+        self.preprocessor = None
+        self.transformed_data = None
+
+        info_box = gui.widgetBox(self.controlArea, "Info")
+        self.input_label = gui.widgetLabel(info_box, "")
+        self.preprocessor_label = gui.widgetLabel(info_box, "")
+        self.output_label = gui.widgetLabel(info_box, "")
+        self.set_input_label_text()
+        self.set_preprocessor_label_text()
+
+    def set_input_label_text(self):
+        text = "No data on input."
+        if self.data is not None:
+            text = "Input data with {:,} instances and {:,} features.".format(
+                len(self.data),
+                len(self.data.domain.attributes))
+        self.input_label.setText(text)
+
+    def set_preprocessor_label_text(self):
+        text = "No preprocessor on input."
+        if self.transformed_data is not None:
+            text = "Preprocessor {} applied.".format(self.preprocessor)
+        elif self.preprocessor is not None:
+            text = "Preprocessor {} on input.".format(self.preprocessor)
+        self.preprocessor_label.setText(text)
+
+    def set_output_label_text(self):
+        text = ""
+        if self.transformed_data:
+            text = "Output data includes {:,} features.".format(
+                len(self.transformed_data.domain.attributes))
+        self.output_label.setText(text)
+
+    @Inputs.data
+    @check_sql_input
+    def set_data(self, data):
+        self.data = data
+        self.set_input_label_text()
+
+    @Inputs.preprocessor
+    def set_preprocessor(self, preprocessor):
+        self.preprocessor = preprocessor
+
+    def handleNewSignals(self):
+        self.apply()
+
+    def apply(self):
+        self.clear_messages()
+        self.transformed_data = None
+        if self.data is not None and self.preprocessor is not None:
+            try:
+                self.transformed_data = self.preprocessor(self.data)
+            except Exception as ex:   # pylint: disable=broad-except
+                self.Error.pp_error(ex)
+        self.Outputs.transformed_data.send(self.transformed_data)
+
+        self.set_preprocessor_label_text()
+        self.set_output_label_text()
+
+    def send_report(self):
+        if self.preprocessor is not None:
+            self.report_items("Settings",
+                              (("Preprocessor", self.preprocessor),))
+        if self.data is not None:
+            self.report_data("Data", self.data)
+        if self.transformed_data is not None:
+            self.report_data("Transformed data", self.transformed_data)
+
+
+if __name__ == "__main__":
+    from AnyQt.QtWidgets import QApplication
+
+    app = QApplication([])
+    ow = OWTransform()
+    d = Table("iris")
+    pp = Discretize()
+    ow.set_data(d)
+    ow.set_preprocessor(pp)
+    ow.handleNewSignals()
+    ow.show()
+    app.exec_()
+    ow.saveSettings()

Orange/widgets/data/owtransform.py

@@ -0,0 +1,92 @@
+# Test methods with long descriptive names can omit docstrings
+# pylint: disable=missing-docstring
+from Orange.data import Table
+from Orange.preprocess import Discretize
+from Orange.preprocess.preprocess import Preprocess
+from Orange.widgets.data.owtransform import OWTransform
+from Orange.widgets.tests.base import WidgetTest
+from Orange.widgets.unsupervised.owpca import OWPCA
+
+
+class TestOWTransform(WidgetTest):
+    def setUp(self):
+        self.widget = self.create_widget(OWTransform)
+        self.data = Table("iris")
+        self.preprocessor = Discretize()
+
+    def test_output(self):
+        # send data and preprocessor
+        self.send_signal(self.widget.Inputs.data, self.data)
+        self.send_signal(self.widget.Inputs.preprocessor, self.preprocessor)
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        self.assertIsInstance(output, Table)
+        self.assertEqual("Input data with 150 instances and 4 features.",
+                         self.widget.input_label.text())
+        self.assertEqual("Preprocessor Discretize() applied.",
+                         self.widget.preprocessor_label.text())
+        self.assertEqual("Output data includes 4 features.",
+                         self.widget.output_label.text())
+
+        # remove preprocessor
+        self.send_signal(self.widget.Inputs.preprocessor, None)
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        self.assertIsNone(output)
+        self.assertEqual("Input data with 150 instances and 4 features.",
+                         self.widget.input_label.text())
+        self.assertEqual("No preprocessor on input.", self.widget.preprocessor_label.text())
+        self.assertEqual("", self.widget.output_label.text())
+
+        # send preprocessor
+        self.send_signal(self.widget.Inputs.preprocessor, self.preprocessor)
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        self.assertIsInstance(output, Table)
+        self.assertEqual("Input data with 150 instances and 4 features.",
+                         self.widget.input_label.text())
+        self.assertEqual("Preprocessor Discretize() applied.",
+                         self.widget.preprocessor_label.text())
+        self.assertEqual("Output data includes 4 features.",
+                         self.widget.output_label.text())
+
+        # remove data
+        self.send_signal(self.widget.Inputs.data, None)
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        self.assertIsNone(output)
+        self.assertEqual("No data on input.", self.widget.input_label.text())
+        self.assertEqual("Preprocessor Discretize() on input.",
+                         self.widget.preprocessor_label.text())
+        self.assertEqual("", self.widget.output_label.text())
+
+        # remove preprocessor
+        self.send_signal(self.widget.Inputs.preprocessor, None)
+        self.assertEqual("No data on input.", self.widget.input_label.text())
+        self.assertEqual("No preprocessor on input.",
+                         self.widget.preprocessor_label.text())
+        self.assertEqual("", self.widget.output_label.text())
+
+    def test_input_pca_preprocessor(self):
+        owpca = self.create_widget(OWPCA)
+        self.send_signal(owpca.Inputs.data, self.data, widget=owpca)
+        owpca.components_spin.setValue(2)
+        pp = self.get_output(owpca.Outputs.preprocessor, widget=owpca)
+        self.assertIsNotNone(pp, Preprocess)
+
+        self.send_signal(self.widget.Inputs.data, self.data)
+        self.send_signal(self.widget.Inputs.preprocessor, pp)
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        self.assertIsInstance(output, Table)
+        self.assertEqual(output.X.shape, (len(self.data), 2))
+
+    def test_error_transforming(self):
+        self.send_signal(self.widget.Inputs.data, self.data)
+        self.send_signal(self.widget.Inputs.preprocessor, Preprocess())
+        self.assertTrue(self.widget.Error.pp_error.is_shown())
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        self.assertIsNone(output)
+        self.send_signal(self.widget.Inputs.data, None)
+        self.assertFalse(self.widget.Error.pp_error.is_shown())
+
+    def test_send_report(self):
+        self.send_signal(self.widget.Inputs.data, self.data)
+        self.widget.report_button.click()
+        self.send_signal(self.widget.Inputs.data, None)
+        self.widget.report_button.click()

Orange/widgets/data/tests/test_owtransform.py

@@ -10,6 +10,7 @@
 from Orange.data import Table, Domain, StringVariable, ContinuousVariable
 from Orange.data.sql.table import SqlTable, AUTO_DL_LIMIT
 from Orange.preprocess import Normalize
+from Orange.preprocess.preprocess import Preprocess, ApplyDomain
 from Orange.projection import PCA, TruncatedSVD
 from Orange.widgets import widget, gui, settings
 from Orange.widgets.widget import Input, Output
@@ -44,6 +45,7 @@ class Outputs:
         transformed_data = Output("Transformed data", Table)
         components = Output("Components", Table)
         pca = Output("PCA", PCA, dynamic=False)
+        preprocessor = Output("Preprocessor", Preprocess)
 
     settingsHandler = settings.DomainContextHandler()
 
@@ -290,6 +292,7 @@ def clear_outputs(self):
         self.Outputs.transformed_data.send(None)
         self.Outputs.components.send(None)
         self.Outputs.pca.send(self._pca_projector)
+        self.Outputs.preprocessor.send(None)
 
     def get_model(self):
         if self.rpca is None:
@@ -455,7 +458,7 @@ def _update_axis(self):
         axis.setTicks([[(i, str(i+1)) for i in range(0, p, d)]])
 
     def commit(self):
-        transformed = components = None
+        transformed = components = pp = None
         if self._pca is not None:
             if self._transformed is None:
                 # Compute the full transform (MAX_COMPONENTS components) only once.
@@ -479,10 +482,13 @@ def commit(self):
                                metas=metas)
             components.name = 'components'
 
+            pp = ApplyDomain(domain, "PCA")
+
         self._pca_projector.component = self.ncomponents
         self.Outputs.transformed_data.send(transformed)
         self.Outputs.components.send(components)
         self.Outputs.pca.send(self._pca_projector)
+        self.Outputs.preprocessor.send(pp)
 
     def send_report(self):
         if self.data is None:

Orange/widgets/unsupervised/owpca.py

@@ -4,6 +4,7 @@
 import scipy.sparse as sp
 
 from Orange.data import Table, Domain, ContinuousVariable, TimeVariable
+from Orange.preprocess.preprocess import Preprocess
 from Orange.widgets.tests.base import WidgetTest
 from Orange.widgets.unsupervised.owpca import OWPCA, DECOMPOSITIONS
 
@@ -131,3 +132,16 @@ def test_do_not_mask_features(self):
         self.widget.set_data(data)
         ndata = Table("iris.tab")
         self.assertEqual(data.domain[0], ndata.domain[0])
+
+    def test_output_preprocessor(self):
+        data = Table("iris")
+        self.send_signal(self.widget.Inputs.data, data)
+        pp = self.get_output(self.widget.Outputs.preprocessor)
+        self.assertIsInstance(pp, Preprocess)
+        transformed_data = pp(data[::10])
+        self.assertIsInstance(transformed_data, Table)
+        self.assertEqual(transformed_data.X.shape, (15, 2))
+        output = self.get_output(self.widget.Outputs.transformed_data)
+        np.testing.assert_array_equal(transformed_data.X, output.X[::10])
+        self.assertEqual([a.name for a in transformed_data.domain.attributes],
+                         [m.name for m in output.domain.attributes])