基于物品的相似度计算

基于物品的协同过滤

import math
import operator

class ItemBasedCF:
    def __init__(self, train_file):
        self.train_file = train_file
        self.readData()

    def readData(self):
        # 读取文件，并生成用户-物品的评分
        # 表和测试集
        self.train = dict()  # 用户-物品的评分表

        for line in open(self.train_file):
            #user,item,score = line.strip().split(",")
            item,user, score  = line.strip().split(",")
            #字典 setdefault() 函数和get() 方法类似, 如果键不存在于字典中，将会添加键并将值设为默认值。
            self.train.setdefault(user, {}) #输出为这种格式{'\xe7\xa5\x9e\xe8\xaf\x9d': {}, '\xe7\x8a\xaf\xe7\xbd\xaa': {}}
            self.train[user][item] = int(float(score))

    def ItemSimilarity(self):
        fw = open('resultData/recomm.txt', 'w')
        # 建立物品-物品的共现矩阵
        C = dict()  # 物品-物品的共现矩阵
        N = dict()  # 物品被多少个不同用户购买
        for user, items in self.train.items():
            for i in items.keys():
                N.setdefault(i, 0)
                N[i] += 1
                C.setdefault(i, {})
                for j in items.keys():
                    if i == j: continue
                    C[i].setdefault(j, 0)
                    C[i][j] += 1
                    # 计算相似度矩阵
        self.W = dict()
        for i, related_items in C.items():
            self.W.setdefault(i, {})
            for j, cij in related_items.items():
                self.W[i][j] = cij / (math.sqrt(N[i] * N[j]))
            remcomdict = {}
            listitems = list(set(self.W.get(i).values()))
            listitems.sort(reverse=True)

            for key, value in sorted(self.W.get(i).items(), key=operator.itemgetter(1), reverse=True):
                if value == listitems[0]:
                    remcomdict[key] = value
                else:
                    break
            fw.writelines(i + ':')
            fw.write(str(remcomdict))
            fw.write('\n')

        fw.close()



Item = ItemBasedCF("data/tvresult.txt")
Item.ItemSimilarity()

格式为：

参考：http://blog.csdn.net/gamer_gyt/article/details/51346159