""" ORM外部调用的配置： """ import numpy as np import math import os def orm_standby(): os.environ.setdefault("DJANGO_SETTINGS_MODULE", "TravelRecommend.settings") # manage.py文件中有同样的环境配置 import django django.setup() def make(data,n): # 二值化处理 n为长度 rea_data = np.zeros(n,dtype=np.int) # 将所选号码填充为1 rea_data[data] = 1 return rea_data # 这里传进去的已经是二值化数据 def calculation(mine,other): # 分子 fenzi = 0 for x in range(len(mine)): fenzi = fenzi + mine[x] * other[x] # 分母 第一部分 fenmu1 = 0 for x in range(len(mine)): fenmu1 = fenmu1 + mine[x] * mine[x] # 给分母 第一部分 开平方 fenmu1 = math.sqrt(fenmu1) # print("分母1开根号", fenmu1) # 分母 第二部分 fenmu2 = 0 for x in range(len(other)): fenmu2 = fenmu2 + other[x] * other[x] # 给分母 第二部分 开平方 fenmu2 = math.sqrt(fenmu2) # print("分母2开根号", fenmu2) # 计算权重 cos_th = fenzi / (fenmu1*fenmu2) # print("********************************推荐符合度",cos_th) return cos_th if __name__ == '__main__': orm_standby() from database import models username = "123" # 推荐用户 user_id = models.UserInfo.objects.get(username=username).pk # 当前用户收藏数组 user_collect_list = [] user_collect_obj = models.Collect.objects.filter(user_id=user_id).values("scenic_spot_id") for collect_id in user_collect_obj: user_collect_list.append(collect_id.get("scenic_spot_id")) # 创建排序组 sort_user_pk_list = [] # 其他用户的收藏数组 collect_list_all = [] user_all = models.UserInfo.objects.values("pk").all() for user_obj in user_all: if user_obj.get("pk") != user_id: sort_user_pk_list.append(user_obj.get("pk")) collect_list_obj_other = models.Collect.objects.filter(user_id=user_obj.get("pk")).values("scenic_spot_id") other_collect_list = [] for collect_id in collect_list_obj_other: other_collect_list.append(collect_id.get("scenic_spot_id")) collect_list_all.append(other_collect_list) # 获取最后一个PK spots_last[0].get("pk") spots_last = models.ScenicSpot.objects.all().order_by("-id").values("pk") # 计算二值化 user_binaryzation_list = make(user_collect_list, int(spots_last[0].get("pk"))) # 计算最推荐值 list_res = [] for data in collect_list_all: list_res.append(calculation(user_binaryzation_list, make(data, int(spots_last[0].get("pk"))))) max_index = list_res.index(max(list_res)) # 筛选出最匹配的用户 print(sort_user_pk_list[max_index])
# 这里做协同过滤 uname = request.COOKIES.get('uname', '')
# 取到当前用户ID user_id_ds = UserInfo.objects.get(uname=uname).pk # 当前用户浏览数组 user_look_list = [] user_look_obj = GoodsBrowser.objects.filter(user_id=user_id_ds).values("good_id") for good_id in user_look_obj: user_look_list.append(good_id.get("good_id")) # 创建排序组 sort_user_pk_list = [] # 其他用户的浏览数组 look_list_all = [] user_all = UserInfo.objects.values("pk").all() for user_obj in user_all: if user_obj.get("pk") != user_id_ds: sort_user_pk_list.append(user_obj.get("pk")) look_list_obj_other = GoodsBrowser.objects.filter(user_id=user_obj.get("pk")).values("good_id") other_look_list = [] for look_id in look_list_obj_other: other_look_list.append(look_id.get("good_id")) look_list_all.append(other_look_list) # print("其他用户的数组",look_list_all) # 获取商品最后一个PK 作为总数 good_count = GoodsInfo.objects.all().order_by("-id").values("pk") print("good_count",int(good_count[0].get("pk"))) # 计算二值化 user_binaryzation_list = make(user_look_list,int(good_count[0].get("pk"))) # 计算最推荐值 list_res = [] for data in look_list_all: list_res.append(calculation(user_binaryzation_list, make(data, int(good_count[0].get("pk"))))) max_index = list_res.index(max(list_res)) # 筛选出最匹配的用户 res_user_id = sort_user_pk_list[max_index]
# 取出改匹配值的用户 取出它的浏览商品 goods_id_set = GoodsBrowser.objects.filter(user_id=res_user_id).values("good_id")[0:2] try: news =[] for good_ids in goods_id_set: res = GoodsInfo.objects.get(pk=good_ids.get("good_id")) news.append(res) except: news = goods.gtype.goodsinfo_set.order_by('-id')[0:2]