增加首页精选的日记点击和美购首页的美购点击

eda/smart_rank/evaluation_metrics.py

@@ -180,11 +180,11 @@ def get_2_tags_coincide_rate(device_order_tags, device_portrait_result, portrait
     return result
 
 
-def get_user_order_info_yesterday():
+def get_user_order_info_yesterday(order_date, order_date_tomorrow):
     # 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
-    sql_order_device_info_yesterday = """
+    sql_device_info_yesterday = """
             SELECT tmp1.user_id,
                    c.device_id,
                    tmp1.service_ids,
@@ -226,25 +226,75 @@ def get_user_order_info_yesterday():
             WHERE c.device_id IS NOT NULL
         """.format(order_date=order_date, order_date_tomorrow=order_date_tomorrow)
     mysql_results = get_data_by_mysql('172.16.30.141', 3306, 'work', 'BJQaT9VzDcuPBqkd', 'zhengxing',
-                                      sql_order_device_info_yesterday)
+                                      sql_device_info_yesterday)
     device_ids_info = [(i["device_id"], int(i["pay_time"])) for i in mysql_results]
-    all_device_order_tags = {i["device_id"]: [int(tag) for tag in i["tag_ids"].split(",")] for i in mysql_results}
-    return device_ids_info, all_device_order_tags
-
-
-def get_user_diary_click_info_yesterday():
-    pass
-
-
-def get_user_service_click_info_yesterday():
-    pass
+    all_device_action_tags = {i["device_id"]: [int(tag) for tag in i["tag_ids"].split(",")] for i in mysql_results}
+    return device_ids_info, all_device_action_tags
+
+
+def get_user_diary_click_info_yesterday(click_date, click_date_tomorrow):
+    # 获取昨天在首页精选点击日记的用户设备id，点击的日记，日记对应的tag
+    # 一个用户对应多个日记，一个日记对应多个tag
+    sql_device_info_yesterday = """
+        SELECT cl_id device_id,
+               max(click_time) AS click_time,
+               group_concat(DISTINCT `diary_id` separator ',') diary_ids,
+               group_concat(DISTINCT `tag_id` separator ',') tag_ids
+        FROM
+          (SELECT d.cl_id,
+                  d.diary_id,
+                  unix_timestamp(d.click_time) AS click_time,
+                  e.tag_id
+           FROM jerry_test.user_click_diary_log d
+           LEFT JOIN eagle.src_mimas_prod_api_diary_tags e ON d.diary_id = e.diary_id
+           LEFT JOIN eagle.src_zhengxing_api_tag f ON e.tag_id = f.id
+           WHERE d.action_from='home精选'
+             AND d.action='on_click_card'
+             AND d.click_time>'{click_date}'
+             AND d.click_time<'{click_date_tomorrow}'
+             AND f.tag_type+0 <'4'+0) tmp2
+        GROUP BY cl_id
+    """.format(click_date=click_date, click_date_tomorrow=click_date_tomorrow)
+    mysql_results = get_data_by_mysql('172.16.40.158', 4000, 'root', '3SYz54LS9#^9sBvC', 'jerry_test', sql_device_info_yesterday)
+    device_ids_info = [(i["device_id"], int(i["click_time"])) for i in mysql_results]
+    all_device_action_tags = {i["device_id"]: [int(tag) for tag in i["tag_ids"].split(",")] for i in mysql_results}
+    return device_ids_info, all_device_action_tags
+
+
+def get_user_service_click_info_yesterday(click_date, click_date_tomorrow):
+    # 获取昨天在美购首页有过点击的用户设备id，点击的美购，美购对应的tag
+    # 一个用户对应多个美购，一个美购对应多个tag
+    sql_device_info_yesterday = """
+            SELECT cl_id device_id,
+                   max(click_time) AS click_time,
+                   group_concat(DISTINCT `service_id` separator ',') service_ids,
+                   group_concat(DISTINCT `tag_id` separator ',') tag_ids
+            FROM
+              (SELECT d.cl_id,
+                      d.service_id,
+                      unix_timestamp(d.click_time) AS click_time,
+                      e.tag_id
+               FROM jerry_test.user_click_service_log d
+               LEFT JOIN eagle.src_zhengxing_api_servicetag e ON d.service_id = e.service_id
+               LEFT JOIN eagle.src_zhengxing_api_tag f ON e.tag_id = f.id
+               WHERE d.action_from='welfare_home_list_item'
+                 AND d.action='goto_welfare_detail'
+                 AND d.click_time>'{click_date}'
+                 AND d.click_time<'{click_date_tomorrow}'
+                 AND f.tag_type+0 <'4'+0) tmp2
+            GROUP BY cl_id
+        """.format(click_date=click_date, click_date_tomorrow=click_date_tomorrow)
+    mysql_results = get_data_by_mysql('172.16.40.158', 4000, 'root', '3SYz54LS9#^9sBvC', 'jerry_test', sql_device_info_yesterday)
+    device_ids_info = [(i["device_id"], int(i["click_time"])) for i in mysql_results]
+    all_device_action_tags = {i["device_id"]: [int(tag) for tag in i["tag_ids"].split(",")] for i in mysql_results}
+    return device_ids_info, all_device_action_tags
 
 
 if __name__ == '__main__':
     try:
         parser = argparse.ArgumentParser(description='画像匹配度的统计')
         my_yesterday = str(datetime.date.today() - datetime.timedelta(days=1))
-        parser.add_argument("-o", "--order_date", type=str, dest="order_date", default=my_yesterday, help="统计的下单日期")
+        parser.add_argument("-o", "--order_date", type=str, dest="order_date", default=my_yesterday, help="统计的行为日期")
         parser.add_argument("-log1", "--log1_file", type=str, dest="portrait_stat_log_path",
                             default="portrait_stat.log", help="画像统计的日志地址")
         parser.add_argument("-log2", "--log2_file", type=str, dest="debug_portrait_stat_log_path",
@@ -288,24 +338,26 @@ if __name__ == '__main__':
         for action in action_type:
             # 获取昨天产生行为的设备id、以及行为对应的tag
             device_ids_lst = list()
-            all_device_order_tags = dict()
+            all_device_action_tags = dict()
             if "order" in action_type:
-                device_ids_lst, all_device_order_tags = get_user_order_info_yesterday()
+                device_ids_lst, all_device_action_tags = get_user_order_info_yesterday(order_date, order_date_tomorrow)
             elif "diary" in action_type:
-                device_ids_lst, all_device_order_tags = get_user_diary_click_info_yesterday()
+                device_ids_lst, all_device_action_tags = get_user_diary_click_info_yesterday(order_date,
+                                                                                             order_date_tomorrow)
             elif "service" in action_type:
-                device_ids_lst, all_device_order_tags = get_user_service_click_info_yesterday()
+                device_ids_lst, all_device_action_tags = get_user_service_click_info_yesterday(order_date,
+                                                                                               order_date_tomorrow)
             else:
                 break
 
             # tags扩展2级tags
-            all_device_order_tags2 = dict()
-            for device in all_device_order_tags:
-                tags = all_device_order_tags[device]
+            all_device_action_tags2 = dict()
+            for device in all_device_action_tags:
+                tags = all_device_action_tags[device]
                 for tag in tags:
                     tags2 = all_3tag_2tag.get(tag, [])
                     tags += tags2
-                all_device_order_tags2[device] = tags
+                all_device_action_tags2[device] = tags
 
             # 用户的去除支付行为的画像
             all_device_portrait_result = dict()
@@ -325,7 +377,7 @@ if __name__ == '__main__':
                 debug_all_device_portrait_result[device] = debug_portrait_result
 
             # 比较两个tag列表的重合率
-            result = get_2_tags_coincide_rate(all_device_order_tags2, all_device_portrait_result, cmd_portrait_top_n,
+            result = get_2_tags_coincide_rate(all_device_action_tags2, all_device_portrait_result, cmd_portrait_top_n,
                                               cmd_coincide_n)
 
             # 有画像没匹配上的用户的画像信息
@@ -347,7 +399,7 @@ if __name__ == '__main__':
             for device in no_coincide_devices:
                 no_coincide_devices_debug = dict()
                 device_portrait_n = all_device_portrait_result[device][:args.portrait_top_n]
-                device_order_tags = all_device_order_tags2[device]
+                device_order_tags = all_device_action_tags2[device]
                 debug_device_portrait_result = debug_all_device_portrait_result[device]
                 no_coincide_devices_debug[device] = {
                     "画像的前{top_n}个tag".format(top_n=args.portrait_top_n): [debug_device_portrait_result[tag] for tag in