update

eda/smart_rank/evaluation_metrics.py

@@ -84,16 +84,31 @@ def get_user_service_portrait_not_alipay(cl_id, all_word_tags, all_tag_tag_type,
 def get_2_tags_coincide_rate(device_order_tags, device_portrait_result, portrait_top_n, coincide_n):
     device_count = len(device_order_tags)
     coincide_count = 0
+    not_coincide_no_portrait = 0
+    not_coincide_no_portrait_device_ids = []
+    not_coincide_have_portrait_device_ids = []
+    not_coincide_have_portrait = 0
     for device in device_order_tags:
         order_tags = device_order_tags[device]
         portrait_tags = device_portrait_result[device]
         if portrait_tags:
             portrait_tags = portrait_tags[:portrait_top_n]
         else:
-            portrait_tags = []
+            not_coincide_no_portrait += 1
+            not_coincide_no_portrait_device_ids.append(device)
+            continue
         if len(set(order_tags).intersection(set(portrait_tags))) >= coincide_n:
             coincide_count += 1
-    return coincide_count/device_count
+        else:
+            not_coincide_have_portrait += 1
+            not_coincide_have_portrait_device_ids.append(device)
+    coincide_rate = coincide_count/device_count
+    result = {"device_count": device_count, "coincide_count": coincide_count, "coincide_rate": coincide_rate,
+              "not_coincide_have_portrait_count": not_coincide_have_portrait,
+              "not_coincide_no_portrait_count": not_coincide_no_portrait,
+              "not_coincide_no_portrait_device_ids": not_coincide_no_portrait_device_ids,
+              "not_coincide_have_portrait_device_ids": not_coincide_have_portrait_device_ids}
+    return result
 
 
 if __name__ == '__main__':
@@ -154,11 +169,21 @@ if __name__ == '__main__':
         # 昨天下单了的用户的去除支付行为的画像
         all_device_portrait_result = dict()
         for device in device_ids_lst:
-            portrait_result = get_user_service_portrait_not_alipay(device, all_word_tags, all_tag_tag_type, all_3tag_2tag, size=10)
+            portrait_result = get_user_service_portrait_not_alipay(device, all_word_tags, all_tag_tag_type,
+                                                                   all_3tag_2tag, size=10)
             all_device_portrait_result[device] = portrait_result
 
         # 比较两个tag列表的重合率
-        rate = get_2_tags_coincide_rate(all_device_order_tags, all_device_portrait_result, 3, 1)
-        print(rate)
+        cmd_portrait_top_n = sys.argv[1]
+        cmd_coincide_n = sys.argv[2]
+        result = get_2_tags_coincide_rate(all_device_order_tags, all_device_portrait_result, cmd_portrait_top_n,
+                                          cmd_coincide_n)
+        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"])
     except Exception as e:
         print(e)
\ No newline at end of file