generate label for mysql

process_data.py

@@ -8,6 +8,7 @@ import json
 import asyncio
 import argparse
 import time
+import numpy as np
 
 from tqdm import tqdm
 import jieba.analyse
@@ -268,6 +269,10 @@ class UserProcess(object):
 
             df.append(temp[1:])
         df = pd.DataFrame(df, columns=self.user_features)
+        df['ros_30'] = np.where(int(df['user_view_30']) != 0, df['user_return_30'] / df['user_share_30'], np.nan)
+        df['rov_30'] = np.where(int(df['user_view_30']) != 0, df['user_return_30'] / df['user_view_30'], np.nan)
+        df['ros_3'] = np.where(int(df['user_view_3']) != 0, df['user_return_3'] / df['user_share_3'], np.nan)
+        df['rov_3'] = np.where(int(df['user_view_3']) != 0, df['user_return_3'] / df['user_view_3'], np.nan)
         df.to_json(des_path, orient='records')