def run_examples(): import random user = USER_DICT.keys()[random.randint(0, len(USER_DICT) - 1)] print user top = recommendations.top_matches(USER_DICT, user) print top recs = recommendations.get_recommendations(USER_DICT, user)[:10] print recs url = recs[0][1] more_top = recommendations.top_matches(recommendations.transform_prefs(USER_DICT), url) print more_top
def run_examples():
import random
user = USER_DICT.keys()[random.randint(0, len(USER_DICT) - 1)]
print user
top = recommendations.top_matches(USER_DICT, user)
print top
recs = recommendations.get_recommendations(USER_DICT, user)[:10]
print recs
url = recs[0][1]
more_top = recommendations.top_matches(
recommendations.transform_prefs(USER_DICT), url)
print more_top
def __main__():
tags_dict = build_tag_dict(build_tag_list("programming"))
print "Similar Tags"
recs = recommendations.top_matches(tags_dict, "programming", similarity=ex1.sim_tanimato)
pprint.pprint(recs)
best_rec = recs[0][1]
print "Similar Tag, But Not Sharing Tags"
pprint.pprint([tag for tag in tags_dict[best_rec] if tags_dict[best_rec][tag] and not tags_dict["programming"][tag]])
# >>> __main__()
# Similar Tags
# [(0.17391304347826086, u'development'),
# (0.15789473684210525, u'course'),
# (0.14285714285714285, u'java'),
# (0.1, u'coursera'),
# (0.09090909090909091, u'gwt')]
# Similar Tag, But Not Sharing Tags
# [u'http://wheel.readthedocs.org/en/latest/',
# u'http://littlegists.blogspot.co.uk/2012/12/building-simple-nancy-app-from-scratch.html',
# u'http://msdn.microsoft.com/en-us/library/dn155905.aspx',
# u'http://www.amityadav.name/code-analyzers-for-php/',
# u'http://www.babycenter.com/0_20-fun-silly-development-boosting-games-to-play-with-your-ba_1479310.bc?scid=momsbaby_20130618:4&pe=MlVEME9CaXwyMDEzMDYxOA..',
# u'http://edition.cnn.com/2013/06/19/opinion/technology-change-lives-african-women-jamme/?hpt=hp_t5',
# u'https://django-cookie-consent.readthedocs.org/en/latest/',
# u'http://www.bbc.co.uk/news/world-22964022',
# u'https://docs.google.com/presentation/d/1IRHyU7_crIiCjl0Gvue0WY3eY_eYvFQvSfwQouW9368/present#slide=id.gebc26cd7_8_0']
user_path = "cache/delicious_users"
if os.path.exists(user_path ):
users = load(open(user_path,"r"))
else:
print "fetching usernames"
users = get_similar_users(u"python",count = 5)
users.append("xiaohan2012")#add myself,hehe!
dump(users,open(user_path,"w"))
print "user count: %d" %len(users)
prefs_path = "cache/delicious_prefs"
if os.path.exists(prefs_path ):
prefs = load(open(prefs_path ,"r"))
print len(prefs)
else:
print "fetching user post info"
prefs = get_prefs(users)
dump(prefs,open(prefs_path ,"w"))
ratio = [sum(prefs[user].values())/len(prefs[user].values()) for user in prefs]
ratio.sort(reverse = True)
from recommendations import top_matches,get_recommendations
import random
#user = random.choice((prefs.keys()))
print user
user = "******"
print top_matches(user,prefs)
print get_recommendations(user,prefs)
print "It is empty, I think we might have encountered data sparsity problem, haha"
return userDict
def fillItems(userDict):
allItems = {}
for user in userDict:
for _ in range(3):
try:
posts = get_userposts(user)
break
except:
print 'Failed user ' + user + ', retrying...'
time.sleep(4)
for post in posts:
url = post['url']
userDict[user][url] = 1.0
allItems[url] = 1
for ratings in userDict.values():
for item in allItems:
if item not in ratings:
ratings[item] = 0.0
if __name__ == '__main__':
delusers = initializeUserDict('programming')
fillItems(delusers)
user = delusers.keys()[random.randint(0, len(delusers) - 1)]
print user
print top_matches(delusers, user)[0:5]
print get_recommendations(delusers, user)[0:5]
# 欧几里得距离评价
euclid1 = sim.sim_euclid(critics, 'Lisa Rose', 'Gene Seymour')
euclid2 = nsim.sim_euclid(critics, 'Lisa Rose', 'Gene Seymour')
print("欧几里得距离评价:")
print(euclid1)
print(euclid2)
# 皮尔逊相关度评价
pearson1 = sim.sim_pearson(critics, 'Lisa Rose', 'Gene Seymour')
pearson2 = nsim.sim_pearson(critics, 'Lisa Rose', 'Gene Seymour')
print("皮尔逊相关度评价:")
print(pearson1)
print(pearson2)
# 余弦相似度评价
cosine1 = nsim.sim_cosine(critics, 'Lisa Rose', 'Gene Seymour')
print("余弦相似度评价:")
print(cosine1)
# jaccard相似度评价: 这个例子说明了jaccard在此场景下的不精确性，虽然二者所评价的电影集合都相同，但是评分不同的差异被忽略了。
jaccard1 = sim.sim_jaccard(critics, 'Lisa Rose', 'Gene Seymour')
print("jaccard相似度评价:")
print(jaccard1)
# 基于商品的评价
movie_pearson = top_matches(transform_profs(critics),
'Superman Returns',
similarity=sim.sim_pearson,
n=5)
print(movie_pearson)
import recommendations
critics = recommendations.critics
# print(critics['Lisa Rose'])
#
# print("Similarity using Euclidean Distance: \n")
# # Using Euclidean Distance
# for critic in critics:
# print("Similarity with " + critic + " : " + str(recommendations.sim_distance("Lisa Rose", critic, critics)))
#
# print("Similarity using Pearson Correlation: \n")
# # Using Correlation
# for critic in critics:
# print("Similarity with " + critic + " : " + str(recommendations.sim_pearson("Lisa Rose", critic, critics)))
print("Top 5 Matches for Lisa")
print(recommendations.top_matches("Lisa Rose", critics))
print("Recommendations for Toby")
print(recommendations.get_recommendations("Toby", critics))
movies = recommendations.flip_params(critics)
print("If you like superman returns then")
print(recommendations.top_matches("Superman Returns", movies))
def test_top_matches_length(self): limit = random.randint(0, len(data.critics) - 1) matches = recommendations.top_matches(data.critics, pick_a_name(), n=limit) (matches).should.have.length_of(limit)
import recommendations
import deliciousrec
delicious_users = deliciousrec.initUserDict('Python', 10)
delicious_users = deliciousrec.fillItems(delicious_users)
test_user = delicious_users.keys()[0]
print recommendations.top_matches(delicious_users, test_user)
print recommendations.get_recommendatios(delicious_users, test_user)[0:4]
def test_top_matches_length(self):
limit = random.randint(0, len(data.critics) - 1)
matches = recommendations.top_matches(data.critics,
pick_a_name(),
n=limit)
(matches).should.have.length_of(limit)
users = load(open(user_path, "r"))
else:
print "fetching usernames"
users = get_similar_users(u"python", count=5)
users.append("xiaohan2012") #add myself,hehe!
dump(users, open(user_path, "w"))
print "user count: %d" % len(users)
prefs_path = "cache/delicious_prefs"
if os.path.exists(prefs_path):
prefs = load(open(prefs_path, "r"))
print len(prefs)
else:
print "fetching user post info"
prefs = get_prefs(users)
dump(prefs, open(prefs_path, "w"))
ratio = [
sum(prefs[user].values()) / len(prefs[user].values()) for user in prefs
]
ratio.sort(reverse=True)
from recommendations import top_matches, get_recommendations
import random
#user = random.choice((prefs.keys()))
print user
user = "******"
print top_matches(user, prefs)
print get_recommendations(user, prefs)
print "It is empty, I think we might have encountered data sparsity problem, haha"
# Library imports
import recommendations, deliciousrec
# Initialize delicious users
delusers=deliciousrec.initialize_user_dict('programming')
# Add myself to the dataset
delusers['rogerfernandezg']={}
# Fills delicious users with data from delicious
deliciousrec.fill_items(delusers)
# Show recommendations for specific user
user=delusers.keys()[1]
print recommendations.top_matches(delusers,user)[0:10]
url=recommendations.get_recommendations(delusers,user)[0][1]
print recommendations.top_matches(recommendations.transform_prefs(delusers),url)