把jerry_prod 改成jerry_test

ctr.py

@@ -9,7 +9,7 @@ def get_tail8():
       where stat_date='{}' \
       and cid_type='{}' \
       and device_id regexp '8$';".format(stat_date,cid_type)
-    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
     cursor = db.cursor()
     print("开始获取")
     cursor.execute(sql)
@@ -28,7 +28,7 @@ def get_ctr(user_tuple):
               where stat_date='{}' \
               and cid_type='{}' \
               and device_id in {}".format(stat_date, cid_type, user_tuple)
-    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
     cursor = db.cursor()
     print("开始获取")
     cursor.execute(sql)
@@ -53,7 +53,7 @@ def get_tail6():
     pre_list = tuple(eval(df.loc[0,"list"]))
     print(len(pre_list))
     print(pre_list[:2])
-    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
     sql = "select distinct device_id from data_feed_click \
           where stat_date='{}' \
           and cid_type='{}' \

local/utils.py

@@ -184,7 +184,7 @@ class FFMFormatPandas:
         if self.feature_index_ is None:
             self.feature_index_ = dict()
             last_idx = 0
-
+# 下面这个feature包括y，应该不包括。这是个bug
         for col in df.columns:
             vals = df[col].unique()
             for val in vals:

prepareData.py

@@ -6,7 +6,7 @@ import pymysql
 
 def fetch_data(start_date, end_date):
     # 获取点击表里的device_id
-    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
     sql = "select distinct device_id from data_feed_click"
     click_device_id = con_sql(db,sql)[0].values.tolist()
     print("成功获取点击表里的device_id")
@@ -14,7 +14,7 @@ def fetch_data(start_date, end_date):
     sql = "select cid,device_id,time,stat_date from data_feed_click " \
           "where stat_date >= '{0}' and stat_date <= '{1}'".format(start_date, end_date)
     # 因为上面的db已经关了，需要再写一遍
-    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
     click = con_sql(db,sql)
     click = click.rename(columns={0: "cid", 1: "device_id", 2: "time_date", 3: "stat_date"})
     print("成功获取点击表里的数据")
@@ -28,7 +28,7 @@ def fetch_data(start_date, end_date):
           "where stat_date >= '{0}' and stat_date <= '{1}'".format(start_date, end_date)
     start = time.time()
     # 因为上面的db已经关了，需要再写一遍
-    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
     exposure = con_sql(db,sql)
     end = time.time()
     print("获取曝光表耗时{}分".format((end-start)/60))