Evaluation results combined

StarrySkyrs · Jun 27, 2022 · 23c7e31 · 23c7e31
1 parent 03c4afa
commit 23c7e31
Show file tree

Hide file tree

Showing 4 changed files with 2,645 additions and 7 deletions.
diff --git a/code/NER_Model.py b/code/NER_Model.py
@@ -350,15 +350,18 @@ def predict(sentence, correct_spellings=True):
 
     output_str = crf.predict([feat])
     print(f" IOB Tagging : {output_str}")
-    output_str = NER_Model_postprocessing.tag_color(output_str)
-    output_str = NER_Model_postprocessing.tag_price(output_str)
-    output_str = NER_Model_postprocessing.tag_shape(output_str)
-    output_str = NER_Model_postprocessing.tag_size(output_str)
-
-    print(f"Postprocessed output: {output_str}")
     result = []
+
     for token, tag in zip(tokens, output_str[0]):
         result.append((token, tag))
+
+    result = NER_Model_postprocessing.tag_color(result)
+    result = NER_Model_postprocessing.tag_price(result)
+    result = NER_Model_postprocessing.tag_shape(result)
+    result = NER_Model_postprocessing.tag_size(result)
+
+    print(f"Postprocessed output: {result}")
+
     return result
 
 def evaluate_results(token_list):

diff --git a/code/evaluation.py b/code/evaluation.py
@@ -200,6 +200,7 @@ def evaluate(true_seqs, pred_seqs, verbose=True):
 tokens = []
 actual_tags = []
 predicted_tags = []
+evaluation_res = {"Token":[], "Actual Tag":[], "Predicted Tag":[]}
 NER_Model.main()
 for index, row in validataion_df.iterrows():
     if row['Tokens'] is np.nan:
@@ -209,18 +210,24 @@ def evaluate(true_seqs, pred_seqs, verbose=True):
         pred_tok_tag = Negation_Analyser.predict(predicted_ner, "tags")
         for tag in pred_tok_tag['token-tags']:
             predicted_tags.append(tag[1])
+            evaluation_res["Predicted Tag"].append(tag[1])
         print(f"Predicted : {pred_tok_tag}")
         sent = []
         tok_tag_act = []
 
     else:
         sent.append(row['Tokens'])
+        evaluation_res["Token"].append(row['Tokens'])
         if(row['is_negative'] is True):
             tok_tag_act.append((row['Tokens'], row['Tags'].replace('I-', 'I-N-')))
             actual_tags.append(row['Tags'].replace('I-', 'I-N-'))
+            evaluation_res["Actual Tag"].append(row['Tags'].replace('I-', 'I-N-'))
         else:
             actual_tags.append(row['Tags'])    
             tok_tag_act.append((row['Tokens'], row['Tags']))
+            evaluation_res["Actual Tag"].append(row['Tags'])
 
 
-evaluate(actual_tags, predicted_tags)
+evaluate(actual_tags, predicted_tags)
+res_df = pd.DataFrame(evaluation_res)
+res_df.to_csv('data/combined_evaluation.csv',index=False)