add logging

eda/smart_rank/evaluation_metrics.py

@@ -8,6 +8,7 @@ import json
 import numpy as np
 import pandas as pd
 from tool import *
+import logging
 
 
 def get_user_service_portrait_not_alipay(cl_id, all_word_tags, all_tag_tag_type, all_3tag_2tag, size=10):
@@ -116,6 +117,7 @@ if __name__ == '__main__':
         # 获取昨天下单的用户设备id，下单的美购，美购对应的tag
         # api_order只有用户的user_id，一个user_id对应多个device_id
         # 用户一次可以下多个订单(美购)，一个美购对应多个tag
+        LOG_DIR = "/home/gmuser/gyz/log/"
         my_today = str(datetime.date.today())
         my_yesterday = str(datetime.date.today() - datetime.timedelta(days=1))
         sql_order_device_info_yesterday = """
@@ -191,7 +193,7 @@ if __name__ == '__main__':
         # 统计画像更新的耗时和更新的设备数
         sql = "select count(*) from user_service_portrait_tags where stat_date='{my_today}'".format(my_today=my_today)
         portrait_device_count = get_data_by_mysql('172.16.40.158', 4000, 'root', '3SYz54LS9#^9sBvC', 'jerry_test', sql)
-        with open("/home/gmuser/gyz/log/dist_portrait.log", 'r') as f:
+        with open(LOG_DIR + "dist_portrait.log", 'r') as f:
             lines = f.readlines()
             start_datetime_str = lines[0][:19]
             end_datetime_str = lines[-1][:19]
@@ -199,18 +201,20 @@ if __name__ == '__main__':
         end_datetime = datetime.datetime.strptime(end_datetime_str, '%Y-%m-%d %H:%M:%S')
         time_consuming = (end_datetime - start_datetime).seconds / 60
 
-        print("画像信息统计日期:", my_today)
-        print("画像更新耗时:{time_consuming}分钟".format(time_consuming=time_consuming))
-        print("画像更新的设备数:", portrait_device_count[0]["count(*)"])
-        print("")
-        print("统计画像匹配度所用数据的日期:", my_yesterday)
-        print("画像的前n个tag:", cmd_portrait_top_n)
-        print("重合个数:", cmd_coincide_n)
-        print("下单人数:", result["device_count"])
-        print("比对的上的人数:", result["coincide_count"])
-        print("匹配度:", result["coincide_rate"])
-        print("比对不上的有画像的人数:", result["not_coincide_have_portrait_count"])
-        print("比对不上的无画像的人数:", result["not_coincide_no_portrait_count"])
-        print("="*100)
+        logging.basicConfig(format='%(message)s', filename=LOG_DIR + 'portrait_stat.log', filemode='a',
+                            level=logging.INFO)
+        logging.info({"画像信息统计日期": my_today})
+        logging.info({"画像更新耗时(分钟)": time_consuming})
+        logging.info({"画像更新的设备数": portrait_device_count[0]["count(*)"]})
+        logging.info("")
+        logging.info({"统计画像匹配度所用数据的日期": my_yesterday})
+        logging.info({"统计画像的选取前n个tag": cmd_portrait_top_n})
+        logging.info({"重合个数": cmd_coincide_n})
+        logging.info({"下单人数": result["device_count"]})
+        logging.info({"比对的上的人数": result["coincide_count"]})
+        logging.info({"匹配度": result["coincide_rate"]})
+        logging.info({"比对不上的有画像的人数": result["not_coincide_have_portrait_count"]})
+        logging.info({"比对不上的无画像的人数": result["not_coincide_no_portrait_count"]})
+        logging.info("="*66)
     except Exception as e:
         print(e)
\ No newline at end of file