def recommendPosts(dataModel):
model = MatrixPreferenceDataModel(dataModel)
print 'User ID`s: '
print model.user_ids()
## print 'Item ID`s: '
## print model.item_ids()
userID= input('Please enter a userID: ')
print 'Loading recommended posts...'
similarity = UserSimilarity(model, pearson_correlation)
recommender = UserBasedRecommender(model, similarity, with_preference=True)
return recommender.recommend(userID)
def recom(i):
try:
M = pd.read_pickle(path + 'DA_Rating.df')
user = M.ix[:, i]
user.name
usernnz = user > 0
usernnzIDX = np.where(usernnz)
targets = user[usernnz].index.values
N = M.copy()
N.ix[:, i] = 0
idx = np.random.random_integers(
0, len(usernnzIDX[0]), min(len(usernnzIDX[0]) - 3, numFeedbacks))
N.ix[idx, i] = user.iloc[idx]
model = MatrixPreferenceDataModel(getRecommenderData(N))
similarity = UserSimilarity(model, cosine_distances, 3)
recommender = UserBasedRecommender(model,
similarity,
with_preference=True)
rec = recommender.recommend(user.name, how_many=100)
ranking = map(lambda x: x[0], rec)
ap, mrr = AP(ranking, targets), MRR(ranking, targets)
pd.to_pickle((ap, mrr),
outpath + 'numFB{}i{}.pkl'.format(numFeedbacks, i))
except:
pass
def build_and_save_recommendations(self, user_items_preference_map):
model = MatrixPreferenceDataModel(user_items_preference_map)
#item_similarity = ItemSimilarity(model, loglikehood_coefficient)
user_similarity = UserSimilarity(model, loglikehood_coefficient)
item_recommendations = ItemBasedColaborativeFiltering(user_items_preference_map).build_recommendations()
user_recommender = UserBasedRecommender(model, user_similarity, with_preference=True)
for item, item_weight_map in item_recommendations.iteritems():
ItemRecommendations.save_recommendations_for_item(item, item_weight_map.iteritems())
user_work_queue = multiprocessing.Queue()
for user_id in model.user_ids().tolist():
user_work_queue.put(user_id)
ParallelWork(UserRecommenderWorker, (user_work_queue, user_recommender)).begin()
def get_recommend(request):
result_list = list()
if 1 == 1:
#判斷是否登入，沒有登入則return not login
if request.user.is_authenticated():
user_id = request.user.id
if user_id == 1:
result_list = [{'result': 'no jimmy'}]
return HttpResponse(json.dumps(result_list))
else:
result_list = [{'result': 'notlogin'}]
return HttpResponse(json.dumps(result_list))
#取回所有評分資料
for row in course_score.objects.all().order_by('u_id'):
temp_dict = {
'user_id': row.u_id,
'course_id': row.c_id.id,
'score': row.cs_score,
'comment': row.cs_comment
}
result_list.append(temp_dict)
#將評分資料做成crab格式
crabdata = getCrabFormat(result_list)
# return HttpResponse((result_list))
#推薦系統
model = MatrixPreferenceDataModel(crabdata)
similarity = UserSimilarity(model, pearson_correlation)
recommender = UserBasedRecommender(model,
similarity,
with_preference=True)
recommend_list = recommender.recommend(user_id)
# return HttpResponse((recommend_list))
recommend_list = [
list(recommend_list[i]) for i in range(0, len(recommend_list))
]
#推薦後資料加上課程名稱、課程資訊...
result_list = list()
for i in range(0, len(recommend_list)):
for course_row in course.objects.filter(id=recommend_list[i][0]):
tem_dict = {
'name': course_row.c_name,
'teacher': course_row.c_teacher,
'credit': course_row.raw_course_set.all()[0].rc_credit,
'depart': course_row.raw_course_set.all()[0].rc_department,
'semester': course_row.raw_course_set.all()[0].rc_semester,
'c_id': recommend_list[i][0],
'raw_cid': course_row.raw_course_set.all()[0].id,
'score': recommend_list[i][1]
}
result_list.append(tem_dict)
return HttpResponse(json.dumps(result_list))
else:
result_list = [{'result': 'error'}]
return HttpResponse(json.dumps(result_list))
def data_points_to_crab_model(data_points):
from scikits.crab.models import MatrixPreferenceDataModel
from collections import defaultdict
data = defaultdict(dict)
for (u, i, r) in data_points:
data[u][i] = r
model = MatrixPreferenceDataModel(data)
return model
def custom_recommend(dataset, recommend_to, num_of_recommendations = 10): #This is based on the USER-BASED RECOMMENDATION Model #Build the model. It returns a dictionary of the recommendations model = MatrixPreferenceDataModel(dataset) similarity = UserSimilarity(model, euclidean_distances) #Build the Recommender recommender = UserBasedRecommender(model, similarity, with_preference = True) return recommender.recommend(recommend_to, num_of_recommendations)
def api_import1():
movies = datasets.load_sample_movies()
import pprint ## to make printed items clearer
pprint.pprint(movies.data)
pprint.pprint(movies.user_ids)
pprint.pprint(movies.item_ids)
from scikits.crab.models import MatrixPreferenceDataModel
#Build the model
model = MatrixPreferenceDataModel(movies.data)
from scikits.crab.metrics import pearson_correlation
from scikits.crab.similarities import UserSimilarity
#Build the similarity
similarity = UserSimilarity(model, pearson_correlation)
from sklearn.base import BaseEstimator
from scikits.crab.recommenders.knn import UserBasedRecommender
#build the user Based recommender
recommender = UserBasedRecommender(model, similarity, with_preference=True)
#recommend item for the user 5 (Toby)
recomendations = recommender.recommend(5)
pprint.pprint(recomendations)
return 'List of ' + url_for('api_articles')
def user_base(input_file, output_file, data_file):
# 基础数据-测试数据
from scikits.crab import datasets
# movies = datasets.load_sample_movies()
user_name = read_user_name(input_file)
shopping_history = load_test_data(data_file)
user_order_num = userName_to_userOrderNum(data_file, user_name)
#Build the model
from scikits.crab.models import MatrixPreferenceDataModel
model = MatrixPreferenceDataModel(shopping_history.data)
#Build the similarity
# 选用算法 pearson_correlation
# from scikits.crab.metrics import pearson_correlation # need score
from scikits.crab.metrics import cosine_distances # DO NOT need score
from scikits.crab.similarities import UserSimilarity
similarity = UserSimilarity(model, cosine_distances)
# 选择 基于User的推荐
from scikits.crab.recommenders.knn import UserBasedRecommender
recommender = UserBasedRecommender(model, similarity, with_preference=True)
ret = recommender.recommend(
user_order_num) # 输出个结果看看效果 Recommend items for the user 5 (Toby)
if ret:
# 所有推荐结果都保存到文件，交给服务器处理
f_w = open(output_file, 'w')
for i in ret:
product_order_num = i[0]
product_score = i[1]
product_name = productOrderNum_to_productName(
data_file, product_order_num)
f_w.write(str(product_name) + ',' + str(product_score) + '\n')
from scikits.crab import datasets movies = datasets.load_sample_movies() songs = datasets.load_sample_songs() print movies.data print movies.user_ids print movies.item_ids from scikits.crab.models import MatrixPreferenceDataModel #Build the model model = MatrixPreferenceDataModel(movies.data) from scikits.crab.metrics import pearson_correlation from scikits.crab.similarities import UserSimilarity #Build the similarity similarity = UserSimilarity(model, pearson_correlation) print similarity[1] from scikits.crab.metrics.pairwise import euclidean_distances similarity = UserSimilarity(model, euclidean_distances) print similarity[5] from scikits.crab.recommenders.knn import UserBasedRecommender #Build the User based recommender recommender = UserBasedRecommender(model, similarity, with_preference=True) #Recommend items for the user 5 (Toby) recommender.recommend(5)
from scikits.crab.models import MatrixPreferenceDataModel
from scikits.crab.metrics import pearson_correlation
from scikits.crab.similarities import UserSimilarity
from pymongo import MongoClient
import os
# get data from the database
client = MongoClient('mongodb://localhost:27017/')
db = client['if']
# TODO generate a table of user bubble ratings
# can be based on number of visits or something
data = db['collaborativeFilterData'].find()
model = MatrixPreferenceDataModel(data)
# backup the old model and save the new one
os.rename("/data/models/collaborativeFilter",
"/data/models/collaborativeFilter_bk")
model.save("/data/models/collaborativeFilter")
def collaborativeFilter(userId, searchTerm):
data = os.read("/data/models/collaborativeFilter")
model = MatrixPreferenceDataModel(data)
similarity = UserSimilarity(model, pearson_correlation)
recommender = UserBasedRecommender(model, similarity, with_preference=True)
return recommender.recommend(50)
for (item, pre) in pre:
# print 'item=', item, ',pre=', pre
if np.isnan(pre):
print 'item=', item
def testParam(how_many=None, **param):
print 'how_many:', how_many
print 'param:', param
if __name__ == '__main__':
# testParam(how_many=12, tt=10, dd=40)
start_time = time.time()
file_name = 'score_data.txt'
with open(file_name) as f:
d = f.read()
# print 'd=', d
d = eval(d)
model = MatrixPreferenceDataModel(d) #build the model
similarity = UserSimilarity(model, pearson_correlation, num_best=50) #build the similarity
recommender = UserBasedRecommender(model, similarity, with_preference=True) #build the recommender
print recommender.recommend(user_id=31071, how_many=30, minimal_similarity=0.8)
# print recommender.recommend(20832)
# print recommender.most_similar_users(user_id=31071)
# print d[20832]
print '====recommand time:', time.time() - start_time
'''
for key in d:
print '===key===', key
print recommender.recommend(key)
'''
# search the keyword in recent and popular tweets
twitter = Twython(APP_KEY, APP_SECRET, OAUTH_TOKEN, OAUTH_TOKEN_SECRET)
tweets = twitter.search(q=keyword, count=100, result_type='mixed')
# count the text's words
from collections import Counter
user_pref_words_list = {}
count = 0
for u_num in range(0, len(tweets['statuses'])):
text = tweets['statuses'][u_num]['text']
user_pref_words_list[u_num] = dict(
Counter(text.replace('\n', ' ').split(' ')))
# calculate user's preference based on frequency of the number of words by using crab
from scikits.crab.models import MatrixPreferenceDataModel
model = MatrixPreferenceDataModel(user_pref_words_list)
from scikits.crab.metrics import pearson_correlation
from scikits.crab.similarities import UserSimilarity
similarity = UserSimilarity(model, pearson_correlation)
from scikits.crab.recommenders.knn import UserBasedRecommender
recommender = UserBasedRecommender(model, similarity, with_preference=True)
# calculate each word preference average
from collections import defaultdict
words_pref = defaultdict(float)
for u_num, pref_words_list in user_pref_words_list.items():
for word, count in pref_words_list.items():
if (words_pref[word]):
words_pref[word] = (words_pref[word] + float(count)) / 2.0
7: {
1: 2.5,
4: 3.5,
5: 4.0
}
}
movies['user_ids'] = {
1: 'Jack Matthews',
2: 'Mick LaSalle',
3: 'Claudia Puig',
4: 'Lisa Rose',
5: 'Toby',
6: 'Gene Seymour',
7: 'Michael Phillips'
}
movies['item_ids'] = {
1: 'Lady in the Water',
2: 'Snakes on a Planet',
3: 'You, Me and Dupree',
4: 'Superman Returns',
5: 'The Night Listener',
6: 'Just My Luck'
}
model = MatrixPreferenceDataModel(movies['data'])
similarity = UserSimilarity(model, pearson_correlation)
recommender = UserBasedRecommender(model, similarity, with_preference=True)
print recommender.recommend(5)
def recommender(request):
freelancer = {"user_ids": {}, 'data': {}, 'item_ids': {}}
for user in User.objects.all():
freelancer['user_ids'][user.id] = user.username
for user in User.objects.all():
freelancer['item_ids'][user.id] = user.username
test_item = {"ios": 1, "ui": 2, "android": 3, "website": 4}
for user in User.objects.all():
freelancer['data'][user.id] = {}
for project in Project.objects.all():
if project.user_id.username == project.user_id.id:
freelancer['data'][user.id][test_item[proj_type]] = 4
break
for id, item in test_item.items():
freelancer['data'][user.id][item] = random.randint(1, 4)
print freelancer['data']
freelancer['data'] = {
1: {
1: 3.0,
2: 4.0,
3: 3.5,
4: 5.0,
5: 3.0
},
2: {
1: 3.0,
2: 4.0,
3: 2.0,
5: 3.0,
6: 2.0
},
3: {
2: 3.5,
3: 2.5,
4: 4.0,
5: 4.5,
6: 3.0
},
4: {
1: 2.5,
2: 3.5,
3: 2.5,
4: 3.5,
5: 3.0,
6: 3.0
},
5: {
2: 4.5,
3: 1.0,
4: 4.0
},
6: {
1: 3.0,
2: 3.5,
3: 3.5,
4: 5.0,
5: 3.0,
6: 1.5
},
7: {
1: 2.5,
4: 3.5,
5: 4.0
}
}
model = MatrixPreferenceDataModel(freelancer['data'])
similarity = UserSimilarity(model, pearson_correlation)
recommender = UserBasedRecommender(model, similarity, with_preference=True)
print freelancer['data']
print recommender.recommend(request.user.id)
return JsonResponse({
"status": True,
"data": recommender.recommend(request.user.id)
})
def create_ml_model_for_recomendations(data):
from scikits.crab.models import MatrixPreferenceDataModel
model = MatrixPreferenceDataModel(data)
return model
from scikits.crab.similarities import UserSimilarity
from scikits.crab.metrics import euclidean_distances
from scikits.crab.recommenders.knn import UserBasedRecommender
# For reproducibility
np.random.seed(1000)
if __name__ == '__main__':
# Define a user-item matrix
user_item_matrix = {
1: {1: 2, 2: 5, 3: 3},
2: {1: 5, 4: 2},
3: {2: 3, 4: 5, 3: 2},
4: {3: 5, 5: 1},
5: {1: 3, 2: 3, 4: 1, 5: 3}
}
# Build a matrix preference model
model = MatrixPreferenceDataModel(user_item_matrix)
# Build a similarity matrix
similarity_matrix = UserSimilarity(model, euclidean_distances)
# Create a recommender
recommender = UserBasedRecommender(model, similarity_matrix, with_preference=True)
# Test the recommender for user 2
with warnings.catch_warnings():
warnings.simplefilter("ignore")
print(recommender.recommend(2))
description='Welcome to Recommender system')
parser.add_argument('Active_user',
type=int,
help='Insert id of user to recommend :')
parser.add_argument('student_ID',
type=int,
help='Insert student_id of user to recommend :')
args = parser.parse_args()
active_user = int(args.Active_user)
student_ID = int(args.student_ID)
#start_time = timeit.default_timer()
myDataset_course = create_finally_dataset(50, active_user, student_ID)
#pprint(myDataset_course)
new_active_user_id = find_new_user_id(myDataset_course, str(student_ID))
model = MatrixPreferenceDataModel(myDataset_course['data'])
similarity_item = ItemSimilarity(model, pearson_correlation)
neighborhood_item = ItemsNeighborhoodStrategy()
recsys_item = ItemBasedRecommender(model,
similarity_item,
neighborhood_item,
with_preference=True)
#recommend_top_5_item = recsys_item.recommended_because(new_active_user_id,42,how_many=5)
recommend_list_item = recsys_item.recommend(new_active_user_id, how_many=5)
#print("Item : " +recommend_list_to_json(recommend_list_item))
#evaluator = CfEvaluator()
#test_item_a = evaluator.evaluate_on_split(recsys_item,at=4, sampling_ratings=0.5,permutation=False,cv=5)
#pprint(test_item_a)
#elapsed = timeit.default_timer() - start_time
#print elapsed
dataset = {}
with open('dataset-recsys.csv') as myfile:
reader = csv.DictReader(myfile, delimiter=',')
i = 0
for line in reader:
i += 1
if (i == 1):
continue
if (int(line['user_id']) not in dataset):
dataset[int(line['user_id'])] = {}
dataset[int(line['user_id'])][int(line['item_id'])] = float(line['star_rating'])
model = MatrixPreferenceDataModel(dataset)
# User-based Similarity
#similarity = UserSimilarity(model, cosine_distances)
#neighborhood = NearestNeighborsStrategy()
#recsys = UserBasedRecommender(model, similarity, neighborhood)
# Item-based Similarity
similarity = ItemSimilarity(model, cosine_distances)
nhood_strategy = ItemsNeighborhoodStrategy()
recsys = ItemBasedRecommender(model, similarity, nhood_strategy, with_preference=False)
#recsys = MatrixFactorBasedRecommender(model=model, items_selection_strategy=nhood_strategy, n_features=10, n_interations=1)
ratings_pd = pd.DataFrame(
ratings, columns=["user_id", "movie_id", "rating", "timestamp"])
ratings_pd = ratings_pd.drop(["timestamp"], 1)
movies = ratings_pd.pivot(index="movie_id", columns="user_id", values="rating")
i = [
'55', '56', '64', '71', '94', '102', '118', '133', '141', '168', '173',
'196', '1278', '928', '780', '651', '514', '483', '432', '404'
]
movies = movies.loc[i, :]
l = [
'13', '128', '201', '207', '222', '23', '234', '246', '267', '268', '269',
'276', '493', '642', '655', '846', '95', '262', '194', '130', '1'
]
movies = movies.loc[:, l]
movies = movies.astype(float).fillna(0.0)
movie_dict = movies.to_dict()
from scikits.crab.models import MatrixPreferenceDataModel
from scikits.crab.metrics import euclidean_distances
from scikits.crab.similarities import ItemSimilarity
from scikits.crab.recommenders.knn import UserBasedRecommender
model = MatrixPreferenceDataModel(movie_dict)
similarity = ItemSimilarity(model, euclidean_distances)
recommender = UserBasedRecommender(model, similarity, with_preference=True)
"""
print '==========movies data============'
print 'user_ids :', movies.user_ids
print 'user_data:', movies.data
print 'item_ids :', movies.item_ids
print '==========movies data============'
"""
"""
#data format of movies
user_ids : {1: 'Jack Matthews', 2: 'Mick LaSalle', 3: 'Claudia Puig', 4: 'Lisa Rose', 5: 'Toby', 6: 'Gene Seymour', 7: 'Michael Phillips'}
user_data: {
1: {1: 3.0, 2: 4.0, 3: 3.5, 4: 5.0, 5: 3.0 },
2: {1: 3.0, 2: 4.0, 3: 2.0, 4: 3.0, 5: 3.0, 6: 2.0},
3: { 2: 3.5, 3: 2.5, 4: 4.0, 5: 4.5, 6: 3.0},
4: {1: 2.5, 2: 3.5, 3: 2.5, 4: 3.5, 5: 3.0, 6: 3.0},
5: { 2: 4.5, 3: 1.0, 4: 4.0 },
6: {1: 3.0, 2: 3.5, 3: 3.5, 4: 5.0, 5: 3.0, 6: 1.5},
7: {1: 2.5, 2: 3.0, 4: 3.5, 5: 4.0 }}
item_ids : {1: 'Lady in the Water', 2: 'Snakes on a Planet', 3: 'You, Me and Dupree', 4: 'Superman Returns', 5: 'The Night Listener', 6: 'Just My Luck'}
"""
model = MatrixPreferenceDataModel(movies.data) #build the model
similarity = UserSimilarity(model, pearson_correlation) #build the similarity
recommender = UserBasedRecommender(
model, similarity, with_preference=True) #build the recommender
# print recommender.recommend(3)
# print recommender.recommend(1, how_many=None, minimal_similarity=5.0)
print recommender.recommend(1, how_many=100, model__minimal_similarity=0.5)
# print recommender.most_similar_users(1)
print data.user_ids
# {1: 'Jack Matthews',
# 2: 'Mick LaSalle',
# 3: 'Claudia Puig',
# 4: 'Lisa Rose',
# 5: 'Toby',
# 6: 'Gene Seymour',
# 7: 'Michael Phillips'}
print data.item_ids
# {1: 'Lady in the Water',
# 2: 'Snakes on a Planet',
# 3: 'You, Me and Dupree',
# 4: 'Superman Returns',
# 5: 'The Night Listener',
# 6: 'Just My Luck'}
# Build the data model
model = MatrixPreferenceDataModel(data.data)
# Build the similarity
similarity = UserSimilarity(model, pearson_correlation)
# Build the User based recommender
recommender = UserBasedRecommender(model, similarity, with_preference=True)
# Recommend items for the user 5 (Toby)
print recommender.recommend(5, how_many=3)
# [(5, 3.3477895267131013), (1, 2.8572508984333034), (6, 2.4473604699719846)]