Unified dataset empty data values

prepare_dataset.py

@@ -32,6 +32,8 @@ def __main(json_file_name):
             current_col.append(person[key])
         df[key] = pd.Series(current_col)
 
+    df = df.drop(columns=['is_closed', 'deactivated'])
+
     pathname, extension = os.path.splitext(json_file_name)
     filename = pathname.split('/')[-1]
 

src/raw_data.py

@@ -23,28 +23,37 @@ class RawData:
     @staticmethod
     def get_int_st(value, attr):
         if value is None:
-            return -1
+            return ''
         result = value[attr]
         if result is None:
-            return -1
+            return ''
         if not str(result).isnumeric():
             print(f'The value {result} is not a number')
-            return -1
+            return ''
         return result
 
+    @staticmethod
+    def str_to_date(value, str_format):
+        return datetime.strptime(value, str_format).date().strftime('%d.%m.%Y')
+
     @staticmethod
     def get_date_st(value):
         if value is None:
             return ''
         try:
-            return datetime.strptime(value, '%d.%m.%Y').date()
+            return RawData.str_to_date(value, '%d.%m.%Y')
         except ValueError:
             try:
-                return datetime.strptime(value, '%d.%m.%y').date()
+                return RawData.str_to_date(value, '%d.%m.%y')
             except ValueError:
                 print(f'Invalid date {value}')
                 return ''
 
     @staticmethod
     def get_collection_st(collection, function):
-        return list(map(lambda item: function(item), [] if collection is None else collection))
+        if collection is None:
+            return ''
+        result_list = list(map(lambda item: function(item), collection))
+        if len(result_list) == 0:
+            return ''
+        return result_list