Small bug fix: good pos ngrams in feature_estractor was being called before instantiation.

ease/create.py

@@ -168,7 +168,7 @@ def _extract_features_and_generate_model(essay_set):
 
     predict_classifier, cv_error_classifier = _instantiate_algorithms(algorithm)
 
-    cv_error_results = get_cv_error(cv_error_classifier, features, essay_set._scores)
+    cv_error_results = _get_cv_error(cv_error_classifier, features, essay_set._scores)
 
     try:
         predict_classifier.fit(features, set_scores)
@@ -213,7 +213,7 @@ def _instantiate_algorithms(algorithm):
     return clf, clf2
 
 
-def get_cv_error(classifier, features, scores):
+def _get_cv_error(classifier, features, scores):
     """
     Gets cross validated error for a given classifier, set of features, and scores
 

ease/feature_extractor.py

@@ -47,6 +47,11 @@ class FeatureExtractor(object):
             max_features_pass_2: The maximum number of features we consider on the second pass of vocabulary grooming
 
         """
+
+        self._good_pos_ngrams = self._get_good_pos_ngrams()
+        self._spell_errors_per_character = 0
+        self._grammar_errors_per_character = 0
+
         if hasattr(essay_set, '_type'):
             if essay_set._type == "train":
                 # Finds vocabulary which differentiates good/high scoring essays from bad/low scoring essays.
@@ -99,10 +104,6 @@ class FeatureExtractor(object):
         else:
             raise util_functions.InputError(essay_set, "wrong input. need an essay set object.")
 
-        self._good_pos_ngrams = self._get_good_pos_ngrams()
-        self._spell_errors_per_character = 0
-        self._grammar_errors_per_character = 0
-
     def generate_features(self, essay_set):
         """
         Generates bag of words, length, and prompt features from an essay set object