Merge pull request #1020 from PrimozGodec/topic-change-default

[ENH] Topic modelling - Reorder methods by relevance

orangecontrib/text/widgets/owtopicmodeling.py

@@ -143,13 +143,14 @@ class Outputs:
     want_main_area = True
 
     methods = [
-        (LsiWidget, 'lsi'),
         (LdaWidget, 'lda'),
+        (LsiWidget, 'lsi'),
         (HdpWidget, 'hdp'),
         (NmfWidget, 'nmf')
     ]
 
     # Settings
+    settings_version = 2
     autocommit = settings.Setting(True)
     method_index = settings.Setting(0)
 
@@ -266,6 +267,8 @@ def on_done(self, corpus):
         if self.model.name == "Latent Dirichlet Allocation":
             bound = self.model.model.log_perplexity(infer_ngrams_corpus(corpus))
             self.perplexity = "{:.5f}".format(np.exp2(-bound))
+        else:
+            self.perplexity = "n/a"
         # for small corpora it is slower to use more processes
         # there is no good estimation when multiprocessing is helpful, but it is
         # definitely not helpful for corpora smaller than 100
@@ -299,6 +302,15 @@ def send_topic_by_id(self, topic_id=None):
             self.Outputs.selected_topic.send(
                 self.model.get_topics_table_by_id(topic_id))
 
+    @classmethod
+    def migrate_settings(cls, settings, version=0):
+        if version < 2 and "method_index" in settings:
+            # in version 2 we change the position of first and second method (lsi, lda)
+            # map changes that correct method from the workflow is loaded
+            change = {1: 0, 0: 1}
+            method_idx = settings["method_index"]
+            settings["method_index"] = change.get(method_idx, method_idx)
+
 
 class TopicViewerTreeWidgetItem(QTreeWidgetItem):
     def __init__(self, topic_id, words, weights, parent,

orangecontrib/text/widgets/tests/test_owtopicmodeling.py

@@ -22,7 +22,6 @@ def test_data(self):
         self.assertIsNone(output)
 
     def test_saved_selection(self):
-        self.widget.method_index = 1
         self.send_signal(self.widget.Inputs.corpus, self.corpus)
         self.wait_until_finished()
 
@@ -58,17 +57,41 @@ def test_topic_evaluation(self):
         self.send_signal(self.widget.Inputs.corpus, self.corpus)
         self.wait_until_finished()
 
-        # test LSI
-        self.assertEqual(self.widget.perplexity, "n/a")
-        self.assertNotEqual(self.widget.coherence, "n/a")
-
         # test LDA, which is the only one with log perplexity
+        self.assertNotEqual(self.widget.perplexity, "n/a")
+        self.assertTrue(self.widget.coherence)
+
+        # test LSI
         self.widget.method_index = 1
         self.widget.commit.now()
         self.wait_until_finished()
+        self.assertEqual(self.widget.perplexity, "n/a")
+        self.assertNotEqual(self.widget.coherence, "n/a")
 
-        self.assertNotEqual(self.widget.perplexity, "n/a")
-        self.assertTrue(self.widget.coherence)
+    def test_migrate_settings_transform(self):
+        # 0 used to be LSI in version <2 - it is on index 1 now
+        settings = {"__version__": 1, "method_index": 0}
+        widget = self.create_widget(OWTopicModeling, stored_settings=settings)
+        self.assertEqual(1, widget.method_index)
+        self.assertEqual("Latent Semantic Indexing", widget.model.name)
+
+        # 1 used to be LDA in version <2 - it is on index 0 now
+        settings = {"__version__": 1, "method_index": 1}
+        widget = self.create_widget(OWTopicModeling, stored_settings=settings)
+        self.assertEqual(0, widget.method_index)
+        self.assertEqual("Latent Dirichlet Allocation", widget.model.name)
+
+        # 2 is unchanged - still HDP
+        settings = {"__version__": 1, "method_index": 2}
+        widget = self.create_widget(OWTopicModeling, stored_settings=settings)
+        self.assertEqual(2, widget.method_index)
+        self.assertEqual("Hierarchical Dirichlet Process", widget.model.name)
+
+        # 2 is unchanged - still NMF
+        settings = {"__version__": 1, "method_index": 3}
+        widget = self.create_widget(OWTopicModeling, stored_settings=settings)
+        self.assertEqual(3, widget.method_index)
+        self.assertEqual("Negative Matrix Factorization", widget.model.name)
 
 
 if __name__ == "__main__":