coursera公开课——recommender system作业（第三周）

懒虫锅～～～！！！！
原始数据：

以及用户目前对doc的数据

#coding:utf-8
import csv
import operator
import math

csvfile=file("data1.csv",'rU')
reader=csv.reader(csvfile,dialect='excel')
doc={}
item=[]
for line in reader:
    if reader.line_num==1:
        for i in range(len(line)):  
            item.append(line[i])
    else:
        for i in range(len(item)):
            if i==0:
                doc.setdefault(line[0],{})
            else:
                doc[line[0]].setdefault(item[i],line[i])
csvfile.close()
csvfile=file('userdata.csv','rU')
reader=csv.reader(csvfile,dialect='excel')
User_Current={}
user=[]
for line in reader:
    if reader.line_num==1:
        for i in range(len(line)):
            user.append(line[i])
    else:
        for i in range(len(line)):
            if i>0:
                User_Current.setdefault(user[i],{})
                if line[i]=='':
                    line[i]="0"
                User_Current[user[i]].setdefault(line[0],line[i])
#计算每个attibute的值DF 
def calDF(doc):
    result={}
    for it in item:
        if it != "":    
            result.setdefault(it,{})
            s=0
            for d in doc:
                s=s+int(doc[d][it])
            result[it]=s
    return result
df=calDF(doc)
#计算每个doc有几个attribute
def numattr(doc):
    result={}
    for d in doc:
        s=0
        for i in doc[d]:
            s=s+int(doc[d][i])
        result.setdefault(d,s)
    return result
totalcount=numattr(doc)
#计算userprofile
def UserProfiles(User_Current,doc):
    result={}
    for person in User_Current:
        result.setdefault(person,{})
        for i in item:
            if i!="":
                s=0.0
                for d in User_Current[person]:
#这行缩进有点问题，第一个part和后两个part的userprofile计算不同
                        s=s+1.0/math.sqrt(int(totalcount[d]))*float(doc[d][i])*float(User_Current[person][d])

                result[person].setdefault(i,s)                      
    return result
userprofile=UserProfiles(User_Current,doc)

#计算dotproduct
def dotProduct(person,userprofile,doc):
    result={}
    for d in doc:
        s=0
        for i in doc[d]:
            s=s+int(doc[d][i])*float(userprofile[person][i])
        result.setdefault(d,s*1.0/math.sqrt(int(totalcount[d])))
    return result

dotproducts=dotProduct('User 1',userprofile,doc)
#计算IDF下的dotproduct=profile*doc_vector*IDF
def IDF(person,userprofile,doc):
    result={}
    for d in doc:
        s=0.0
        for i in doc[d]:
            s=s+int(doc[d][i])*float(userprofile[person][i])*1.0/int(df[i])
        result.setdefault(d,s*1.0/math.sqrt(int(totalcount[d])))
    return result

Idf=IDF('User 2',userprofile,doc)
def topn_dotProduct(dotproducts,n=5):
    sorted_x=sorted(dotproducts.iteritems(),key=operator.itemgetter(1),reverse=True)
    for i in range(n):
        print sorted_x[i]
topn_dotProduct(Idf,20)

问题一：

def UserProfiles(User_Current,doc):
    result={}
    for person in User_Current:
        result.setdefault(person,{})
        for i in item:
            if i!="":
                s=0.0
                for d in User_Current[person]:
                    s=s+float(doc[d][i])*float(User_Current[person][d])
                result[person].setdefault(i,s)                      
    return result
userprofile=UserProfiles(User_Current,doc)

def dotProduct(person,userprofile,doc):
    result={}
    for d in doc:
        s=0
        for i in doc[d]:
            s=s+int(doc[d][i])*float(userprofile[person][i])
        result.setdefault(d,s)
    return result

dotproducts=dotProduct('User 1',userprofile,doc)

def topn_dotProduct(dotproducts,n=5):
    sorted_x=sorted(dotproducts.iteritems(),key=operator.itemgetter(1),reverse=True)
    for i in range(n):
        print sorted_x[i]
topn_dotProduct(dotproducts,20)

问题二：

def UserProfiles(User_Current,doc):
    result={}
    for person in User_Current:
        result.setdefault(person,{})
        for i in item:
            if i!="":
                s=0.0
                for d in User_Current[person]:
                    s=s+1.0/math.sqrt(int(totalcount[d]))*float(doc[d][i])*float(User_Current[person][d])
                result[person].setdefault(i,s)                      
    return result
userprofile=UserProfiles(User_Current,doc)

def dotProduct(person,userprofile,doc):
    result={}
    for d in doc:
        s=0
        for i in doc[d]:
            s=s+int(doc[d][i])*float(userprofile[person][i])
        result.setdefault(d,s*1.0/math.sqrt(int(totalcount[d])))
    return result

dotproducts=dotProduct('User 1',userprofile,doc)

def topn_dotProduct(dotproducts,n=5):
    sorted_x=sorted(dotproducts.iteritems(),key=operator.itemgetter(1),reverse=True)
    for i in range(n):
        print sorted_x[i]
topn_dotProduct(dotproducts,20)

～～～～
遇到的一些问题
直接将表格存为.csv文件，运行python时候会报错
_csv.Error: line contains NULL byte
保存形式应该是另存为.csv

python字典排序问题

#-*- encoding=utf-8 -*-  

import operator  
#按字典值排序（默认为升序）  
x = {1:2, 3:4, 4:3, 2:1, 0:0}  
sorted_x = sorted(x.iteritems(), key=operator.itemgetter(1))  
print sorted_x  
#[(0, 0), (2, 1), (1, 2), (4, 3), (3, 4)]  
#如果要降序排序,可以指定reverse=True  
sorted_x = sorted(x.iteritems(), key=operator.itemgetter(1), reverse=True)  
print sorted_x  
#[(3, 4), (4, 3), (1, 2), (2, 1), (0, 0)]  
#或者直接使用list的reverse方法将sorted_x顺序反转  
#sorted_x.reverse()  

#取代方法是,用lambda表达式  
sorted_x = sorted(x.iteritems(), key=lambda x : x[1])  
print sorted_x  
#[(0, 0), (2, 1), (1, 2), (4, 3), (3, 4)]  
sorted_x = sorted(x.iteritems(), key=lambda x : x[1], reverse=True)  
print sorted_x  
#[(3, 4), (4, 3), (1, 2), (2, 1), (0, 0)]  

#包含字典dict的列表list的排序方法与dict的排序类似,如下：  
x = [{'name':'Homer', 'age':39}, {'name':'Bart', 'age':10}]  
sorted_x = sorted(x, key=operator.itemgetter('name'))  
print sorted_x  
#[{'age': 10, 'name': 'Bart'}, {'age': 39, 'name': 'Homer'}]  
sorted_x = sorted(x, key=operator.itemgetter('name'), reverse=True)  
print sorted_x  
#[{'age': 39, 'name': 'Homer'}, {'age': 10, 'name': 'Bart'}]  
sorted_x = sorted(x, key=lambda x : x['name'])  
print sorted_x  
#[{'age': 10, 'name': 'Bart'}, {'age': 39, 'name': 'Homer'}]  
sorted_x = sorted(x, key=lambda x : x['name'], reverse=True)  
print sorted_x  
#[{'age': 39, 'name': 'Homer'}, {'age': 10, 'name': 'Bart'}]