rerun unigram adaboost to confirm results were aligned

AIclassifier.py

@@ -121,9 +121,9 @@ def create_vectors_list(sents, conversion_dict):
         bigrams_vector.append(sent_bigrams_vector)
         mixed_vector.append(sent_mixed_vector)
 
-    #return vectors_list
+    return vectors_list # unigrams
     #return bigrams_vector
-    return mixed_vector
+    #return mixed_vector
 
 # def create_word_embedding(partition):
 
@@ -195,43 +195,43 @@ def create_vectors_list(sents, conversion_dict):
 #train_word_embedding_features = create_word_embedding(sents_train)
 #dev_word_embedding_features = create_word_embedding(sents_dev)
 #test_word_embedding_features = numpy.asarray(create_word_embedding(sents_test))
-#print("Length of the dictionary of word representations:",len(words_to_numbers))
+print("Length of the dictionary of word representations:",len(words_to_numbers))
 # print("Length of the dictionary of word representations:",len(bigrams_to_numbers))
-print("Length of the dictionary of word representations:",len(mixed_to_numbers))
+#print("Length of the dictionary of word representations:",len(mixed_to_numbers))
 
 # FLAG - CHECK IF DICTIONARY IS BUILT CORRECTLY
 #               SHOULD PUNCTUATION BE UNKNOWN? BECAUSE RIGHT NOW IT IS -NOPE, FIXED
 # TO DO: count frequency again?
 # count frequency before and after removing unknown words - ??? - ASK GABRIEL!!
 # checked that it seems ok
 
-# train_vectors_list = create_vectors_list(sents_train, words_to_numbers)
-# dev_vectors_list = create_vectors_list(sents_dev, words_to_numbers)
-# test_vectors_list = create_vectors_list(sents_test, words_to_numbers)
+train_vectors_list = create_vectors_list(sents_train, words_to_numbers)
+dev_vectors_list = create_vectors_list(sents_dev, words_to_numbers)
+test_vectors_list = create_vectors_list(sents_test, words_to_numbers)
 
 # train_vectors_list = create_vectors_list(sents_train, bigrams_to_numbers)
 # dev_vectors_list = create_vectors_list(sents_dev, bigrams_to_numbers)
 # test_vectors_list = create_vectors_list(sents_test, bigrams_to_numbers)
 
-train_vectors_list = create_vectors_list(sents_train, mixed_to_numbers)
-dev_vectors_list = create_vectors_list(sents_dev, mixed_to_numbers)
-test_vectors_list = create_vectors_list(sents_test, mixed_to_numbers)
+# train_vectors_list = create_vectors_list(sents_train, mixed_to_numbers)
+# dev_vectors_list = create_vectors_list(sents_dev, mixed_to_numbers)
+# test_vectors_list = create_vectors_list(sents_test, mixed_to_numbers)
 
 # COUNT STATISTICS - HOW MANY WORDS WERE CONSIDERED UNK, AND HOW MANY OF EACH WORD
 
 # FLAG - CHECK IF SENTENCE REPRESENTATIONS WERE DONE CORRECTLY
 
-# train_matrix_array = format_sentVector_to_SparseMatrix(train_vectors_list, words_to_numbers)
-# dev_matrix_array = format_sentVector_to_SparseMatrix(dev_vectors_list, words_to_numbers)
-# test_matrix_array = format_sentVector_to_SparseMatrix(test_vectors_list, words_to_numbers)
+train_matrix_array = format_sentVector_to_SparseMatrix(train_vectors_list, words_to_numbers)
+dev_matrix_array = format_sentVector_to_SparseMatrix(dev_vectors_list, words_to_numbers)
+test_matrix_array = format_sentVector_to_SparseMatrix(test_vectors_list, words_to_numbers)
 
 # train_matrix_array = format_sentVector_to_SparseMatrix(train_vectors_list, bigrams_to_numbers)
 # dev_matrix_array = format_sentVector_to_SparseMatrix(dev_vectors_list, bigrams_to_numbers)
 # test_matrix_array = format_sentVector_to_SparseMatrix(test_vectors_list, bigrams_to_numbers)
 
-train_matrix_array = format_sentVector_to_SparseMatrix(train_vectors_list, mixed_to_numbers)
-dev_matrix_array = format_sentVector_to_SparseMatrix(dev_vectors_list, mixed_to_numbers)
-test_matrix_array = format_sentVector_to_SparseMatrix(test_vectors_list, mixed_to_numbers)
+# train_matrix_array = format_sentVector_to_SparseMatrix(train_vectors_list, mixed_to_numbers)
+# dev_matrix_array = format_sentVector_to_SparseMatrix(dev_vectors_list, mixed_to_numbers)
+# test_matrix_array = format_sentVector_to_SparseMatrix(test_vectors_list, mixed_to_numbers)
 
 # FLAG - CHECK IF SPARSE MATRIX REPRESENTATION WAS DONE CORRECTLY
 
@@ -283,8 +283,8 @@ def create_vectors_list(sents, conversion_dict):
 importances = model.feature_importances_
 
 features = {}
-# for i,(token,value) in enumerate(zip(words_to_numbers, importances)):
-for i,(token,value) in enumerate(zip(mixed_to_numbers, importances)):
+for i,(token,value) in enumerate(zip(words_to_numbers, importances)):
+#for i,(token,value) in enumerate(zip(mixed_to_numbers, importances)):
 #for i,(token,value) in enumerate(zip(bigrams_to_numbers, importances)): # IMPORTANTO TO CHANGE TO ADEQUATE DICT
    if (value != 0):
        features[token] = value
@@ -324,7 +324,7 @@ def create_vectors_list(sents, conversion_dict):
 os.makedirs(os.path.dirname(path), exist_ok=True)
 with open(path, 'w') as file:
     #print("Performance measures - Unigram Dictionary - MLP Word Embeddings\n", file=file)
-    print("Performance measures - Mixed Dictionary - Adaboost\n", file=file)
+    print("Performance measures - Unigram Dictionary - Adaboost\n", file=file)
 #write_output_stats_file(path, "Dev", dev_labels_primary, predictions, labels)
 write_output_stats_file(path, "Test", test_labels_primary, predictions, labels)
 

output/AI Classifier/1labelPredictionsStatsTest.txt

@@ -1,14 +1,14 @@
-Performance measures - Mixed Dictionary - Adaboost
+Performance measures - Unigram Dictionary - Adaboost
 
 Test set:
 
-Precision macro: 0.246
-Precision Individually: [0.364 0.    0.615 0.    0.25 ]
-Recall macro: 0.289
-Recall Individually: [0.571 0.    0.774 0.    0.1  ]
-F1 Score micro: 0.493
-F1 Score macro: 0.255
-F1 Score weighted: 0.431
-F1 Score Individually: [0.444 0.    0.686 0.    0.143]
+Precision macro: 0.368
+Precision Individually: [0.556 0.    0.551 0.4   0.333]
+Recall macro: 0.302
+Recall Individually: [0.357 0.    0.871 0.182 0.1  ]
+F1 Score micro: 0.522
+F1 Score macro: 0.303
+F1 Score weighted: 0.467
+F1 Score Individually: [0.435 0.    0.675 0.25  0.154]