Eliminate bias by retraining only with train set

See issue coastalcph#32 for more info about the bias

models/tfidf_svm.py

@@ -81,25 +81,34 @@ def add_zero_class(labels):
         # Fixate Validation Split
         split_index = [-1] * len(dataset['train']) + [0] * len(dataset['validation'])
         val_split = PredefinedSplit(test_fold=split_index)
-        gs_clf = GridSearchCV(text_clf, parameters, cv=val_split, n_jobs=32, verbose=4)
+        gs_clf = GridSearchCV(text_clf, parameters, cv=val_split, n_jobs=32, verbose=4, refit = False)
 
         # Pre-process inputs, outputs
-        x_train = get_text(dataset['train']) + get_text(dataset['validation'])
+        x_train = get_text(dataset['train'])
+        x_val = get_text(dataset['validation'])
+        x_train_val = x_train + x_val
+
         if config.task_type == 'multi_label':
             mlb = MultiLabelBinarizer(classes=range(config.n_classes))
             mlb.fit(dataset['train']['labels'])
         else:
             mlb = None
-        y_train = get_labels(dataset['train'], mlb) + get_labels(dataset['validation'], mlb)
+        y_train = get_labels(dataset['train'], mlb)
+        y_val = get_labels(dataset['validation'], mlb)
+        y_train_val = y_train + y_val
 
         # Train classifier
-        gs_clf = gs_clf.fit(x_train, y_train)
+        gs_clf = gs_clf.fit(x_train_val, y_train_val)
 
         # Print best hyper-parameters
         logging.info('Best Parameters:')
         for param_name in sorted(parameters.keys()):
             logging.info("%s: %r" % (param_name, gs_clf.best_params_[param_name]))
-
+
+        # Retrain model with best CV parameters only with train data
+        text_clf.set_params(**gs_clf.best_params_)
+        gs_clf = text_clf.fit(x_train, y_train)
+
         # Report results
         logging.info('VALIDATION RESULTS:')
         y_pred = gs_clf.predict(get_text(dataset['validation']))