Python Slope One

"" "#slope one algorithm is based on" the same thing "between the" scoring difference "of the linear algorithm, predicting the user to the item scoring personalized algorithm #适用商品的更新不频繁, a relatively stable number of items are significantly smaller than the number of users of the scene, depending on the user's behavior log and object preferences related content Advantages: 1 algorithm is simple, 2 can find the user's potential interests #缺点: dependent user behavior, there are cold startup problems and sparsity problems "" "#计算物品之间评分差的平均值, enter the following #items items-user Rating data Items #users user-item rating data users # Averages storage average Classification dictionary type data def buildaveragediffs (items, users, averages): #遍历每条物品-User Rating data for itemId in items: #遍 
            Calendar each other article-user rating data for Otheritemid in items: #物品间的评分偏差 average = 0.0 #两件物品均评过分的用户数 Userratingpaircount = 0 #若为不同的物品项 if itemId!= otheritemid: #遍历每条用户-Object Tasting data for userId in users: #每条数据为用户对物品的评分 userratings = Users[user Id] #当前物品项在用户的评分数据中, and the user also has a rating for other items if itemId in Userratings and Otheritemid in Userr Atings: #两件物品均评过分的用户数加1 userratingpaircount + 1 #
           Scoring deviations for each current item score-Other Items score sum             Average + = (Userratings[otheritemid]-userratings[itemid]) #物品间的评分偏差等于评分偏差和/Two items are rated excessive number of users averages[(itemid,otheritemid)] = Average/userratingpaircount #推荐评分, enter the following #users user-item rating data users #items items-user Rating data #averages rating deviations between items #targetUserId recommended users Targetuserid #targetItemId recommended items Targetitemid #return Predictive rating def
    Suggestedrating (Users, items, averages, Targetuserid, Targetitemid): #预测评分分母 runningratingcount = 0 #预测评分分子 Weightedratingtotal = 0.0 #遍历需要推荐的用户targetUserId对有过评分的物品 for i in Users[targetuserid]: #保存当前物品与要推荐物品targ Etitemid, common scoring user number Ratingcount = Userswhoratedboth (Users, I, Targetitemid) #分子 Weightedratingtotal + = (Users[targetuserid][i]-averages[(Targetitemid, i)]) * Ratingcount #分母 Runningratingcount + = Ratingcou NT #返回预测评分 return weightedratingtotal/runningratingcount #物品itemId1与itemId2共同有多少用户评分 def userswhoratedboth ( Users, ItemId1, itemId2): Count =0 #用户-Goods rating data users for userId in users: #用户对物品itemId1与itemId2都有过评分则计数值加1 if itemId1 in Users[userid

] and ItemId2 in Users[userid]: #加1 count = 1 #返回itemId1与itemId2共同评分用户数 return count 

    if __name__ = = ' __main__ ': items = {' A ': {1:5, 2:3}, ' B ': {1:3, 2:4, 3:2}, ' C ': {1:2, 3:5}}
    Users = {1: {' A ': 5, ' B ': 3, ' C ': 2}, 2: {' A ': 3, ' B ': 4}, 3: {' B ': 2, ' C ': 5}}} #物品间评分偏差初始化
    averages = {} #计算物品之间评分差的平均值 buildaveragediffs (items, users, averages) print averages #输出: Total items, number of users Print ({' ItemCount ': Len (items), ' UserCount ': Len (Users)}) print ("Guess that user A'll rate Item 3=" + STR (Sugg




 Estedrating (Users,items, averages, 3, ' A '))