增加尾号5或6点击率统计

ctr-56.py

@@ -22,11 +22,42 @@ def get_data():
     device_id = tuple(pd.DataFrame(list(result))[0].values.tolist())
     cursor.close()
     print(device_id[0:2])
+
     return device_id
 
 
+def ctr_all():
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_prod')
+    cursor = db.cursor()
+    sql_active = "select count(distinct device_id) from data_feed_exposure where (cid_type = 'diary' or cid_type = 'diary_video') " \
+                 "and device_id regexp'[5|6]$' and stat_date = '{}';".format(date)
+    cursor.execute(sql_active)
+    tail56 = cursor.fetchone()[0]
+    print("当天尾号5或6活跃用户数：")
+    print(tail56)
+    cover = len(set(device_id) & set(tail56))
+    print("当天尾号5或6活跃用户覆盖数：")
+    print(cover)
+    cover_percent = cover / len(set(device_id))
+    print("当天尾号5或6活跃用户覆盖率：")
+    print(cover_percent)
+    sql_click = "select count(cid) from data_feed_click " \
+                "where (cid_type = 'diary' or cid_type = 'diary_video') " \
+                "and stat_date = '{}' and device_id regexp '[5|6]$';".format(date)
+    cursor.execute(sql_click)
+    click = cursor.fetchone()[0]
+    print("点击数：" + str(click))
+    sql_exp = "select count(cid) from data_feed_exposure " \
+              "where (cid_type = 'diary' or cid_type = 'diary_video') and stat_date = '{}' and " \
+              "device_id regexp '[5|6]$'".format(date)
+    cursor.execute(sql_exp)
+    exp = cursor.fetchone()[0]
+    print("曝光数：" + str(exp))
+    print("点击率：" + str(click / exp))
+    return tail56,cover,cover_percent,click,exp,click / exp
+
+
 def ctr():
-    device_id = get_data()
     sql_click = "select count(cid) from data_feed_click " \
               "where (cid_type = 'diary' or cid_type = 'diary_video') " \
                 "and stat_date = '{}' and device_id in {};".format(date,device_id)
@@ -34,14 +65,14 @@ def ctr():
     cursor = db.cursor()
     cursor.execute(sql_click)
     click = cursor.fetchone()[0]
-    print("点击数："+str(click))
+    print("实验用户点击数："+str(click))
     sql_exp = "select count(cid) from data_feed_exposure " \
               "where (cid_type = 'diary' or cid_type = 'diary_video') and stat_date = '{}' and " \
               "device_id in {}".format(date,device_id)
     cursor.execute(sql_exp)
     exp = cursor.fetchone()[0]
-    print("曝光数："+str(exp))
-    print("点击率："+str(click/exp))
+    print("实验用户曝光数："+str(exp))
+    print("实验用户点击率："+str(click/exp))
 
     return click,exp,click/exp
 
@@ -49,10 +80,14 @@ def ctr():
 def rate2file():
     output_path = DIRECTORY_PATH + "56tail.csv"
     with open(output_path,'a+') as f:
-        line = date.replace('-', '')+','+str(temp_data[0])+','+str(temp_data[1])+','+str(temp_data[2])+'\n'
+        line = date.replace('-', '')+','+str(temp_data[0])+','+str(temp_data[1])+','+str(temp_data[2])+\
+               ","+str(data[0])+","+str(data[1])+","+str(data[2])+","+str(data[3])+","+str(data[4])\
+               +","+str(data[5])+'\n'
         f.write(line)
 
 if __name__ == "__main__":
     date = get_yesterday_date()
+    device_id = get_data()
     temp_data = ctr()
+    data = ctr_all()
     rate2file()