pyspark实现FunkSVD电影推荐系统

from pyspark import SparkConf, SparkContext 
from pyspark.mllib.recommendation import ALS, Rating
# 获取所有movie名称和id对应集合  
def movie_dict(file):  
dict = {}  
    with open(file) as f:  
        for line in f:  
            arr= line.split('|')  
            movie_id = int(arr[0])  
            movie_name = str(arr[1])  
            dict[movie_id] = movie_name  
    return dict  
# 转换用户评分数据格式  
def get_rating(str):  
    #arr = str.split('\t') 
    arr = str.split('::')  
    user_id = int(arr[0])  
    movie_id = int(arr[1])  
    user_rating = float(arr[2])  
    return Rating(user_id, movie_id, user_rating)
try:
    sc.stop()
except:
    pass
conf=SparkConf().setMaster('local').setAppName('MovieRec').set("spark.executor.memory", "520m")  
sc = SparkContext(conf=conf)
#加载数据  
#movies = movie_dict('/home/joe/.surprise_data/ml-100k/ml-100k/u.item')  
#sc.broadcast(movies)  
#data = sc.textFile('/home/joe/.surprise_data/ml-100k/ml-100k/u.data')  
data = sc.textFile('/home/joe/.surprise_data/ml-1m/ml-1m/ratings.dat')  
# 转换 (user, product, rating) tuple  
ratings = data.map(get_rating)  
  
# 建立模型  
rank = 10  
iterations = 5  
model = ALS.train(ratings, rank, iterations)  
  
# 对指定用户ID推荐  
userid = 10  
user_ratings = ratings.filter(lambda x: x[0] == userid)  
  
#按得分高低推荐前10电影  
rec_movies=model.recommendProducts(userid, 10)  
print ('\n################################\n')    
print ('recommend movies for userid %d:' % userid)
for i in rec_movies:
    print(i)
#for item in rec_movies:  
    #print ('name:'+movies[item[1]]+'==> score: %.2f' % item[2]) 
print ('\n################################\n' ) 
sc.stop() 

参考: http://strayly.iteye.com/blog/2341855
按照上面链接会出现ValueError: Cannot run multiple SparkContexts at once; existing SparkContext(app=PySparkShell, master=local[*]) created by at /usr/local/lib/python2.7/dist-packages/IPython/utils/py3compat.py:288
然后参考一下链接进行修改
http://blog.csdn.net/dream_an/article/details/51935530
movies 可以不使用
用Spark学习矩阵分解推荐算法：http://www.cnblogs.com/pinard/p/6364932.html

欢迎关注人工智能见闻公众号