上一節是根據物品item的描述屬性進行基於內容的推薦基於內容的電影推薦，本節中還是以電影推薦為例，講解基於內容的協同推薦演算法。

    python程式碼如下

import pandas as pd
useritemdata=pd.read_csv('C:/Users/Ray/Desktop/recommand/ml01/ratings.csv')
useritemdata.head()

itemlist=list(set(useritemdata['movieId'].tolist()))
# print(itemlist)

usercount=len(set(useritemdata['userId'].tolist()))
print(usercount)
 

5

itemAffinity=pd.DataFrame(columns=('item1','item2','score'))
rowcount=0

for ind1 in range(len(itemlist)):
    item1users=useritemdata[useritemdata.movieId==itemlist[ind1]]['userId'].tolist()
#     print(itemlist[ind1],item1users)
    for ind2 in range(ind1,len(itemlist)):
        if (ind1==ind2):
            continue
        item2users=useritemdata[useritemdata.movieId==itemlist[ind2]]['userId'].tolist()
#         print(itemlist[ind2],item2users)
        commentusers=len(set(item1users).intersection(set(item2users)))
        score=commentusers/usercount
        itemAffinity.loc[rowcount]=[itemlist[ind1],itemlist[ind2],score]
        rowcount+=1
        itemAffinity.loc[rowcount]=[itemlist[ind2],itemlist[ind1],score]
        rowcount+=1
itemAffinity.head()
 

def getrecommendation(searchitem):
    recolist=itemAffinity[itemAffinity.item1==searchitem][['item2','score']].sort_values('score',ascending=[0])
    return recolist

searchitem=5001
recolist=getrecommendation(searchitem)
print('recommendations for movieid'+str(searchitem)+'\n',recolist.head(5))