def mycmp(user1, user2):
if len(supp_user(user1)) < len(supp_user(user2)):
return -1
elif len(supp_user(user1)) > len(supp_user(user2)):
return 1
else:
return set_cmp(user1, user2)
def set_cmp(user1, user2):
items_1 = supp_user(user1)
items_2 = supp_user(user2)
while len(items_1) != 0:
if min(items_1) < min(items_2):
return -1
elif min(items_1) > min(items_2):
return 1
else:
min_value = min(items_1)
items_1.remove(min_value)
items_2.remove(min_value)
return 0
def k_corating_slice(sorted_ratings,
myslice,
items_to_keep=None): # [start,end)
logging.info('k corating slice(%s,%s)', myslice.start, myslice.stop)
part_ratings = sorted_ratings[myslice, :]
items_need_to_rate = set()
for record in part_ratings:
items_need_to_rate = items_need_to_rate.union(supp_user(record))
if items_to_keep is None:
for item_id in items_need_to_rate:
for record in part_ratings:
if record[item_id] == unknown_rating:
record[item_id] = adapter(
pd_rating(original_ratings, int(record[-1] - 1),
item_id, web)[0])
else:
for item_id in items_need_to_rate:
if item_id not in items_to_keep:
for record in part_ratings:
record[item_id] = unknown_rating
else:
for record in part_ratings:
if record[item_id] == unknown_rating:
record[item_id] = adapter(
pd_rating(original_ratings, int(record[-1] - 1),
item_id, web)[0])
def generate_aux(user_id):
def random_wrong(start, end, right_value, tp):
if tp == int:
while True:
r = random.randint(start, end)
if r != right_value:
return r
elif tp == float:
while True:
r = random.uniform(start, end)
if r != right_value:
return r
user = original_ratings[user_id, :]
items = list(supp_user(user))
if total < len(items):
total_list = random.sample(items, total)
else:
total_list = items
if correct < len(total_list):
correct_list = random.sample(total_list, correct)
else:
correct_list = total_list
aux = [0] * item_size
for i in total_list:
if i in correct_list:
aux[i] = user[i]
else:
aux[i] = random_wrong(min_rating, max_rating, user[i], int)
return aux
def score(aux, record):
item_ids = supp_user(aux)
sum = 0
for item_id in item_ids:
weight = item_weight(item_id)
sum += weight * sim_rate(aux[item_id], record[item_id])
return sum
def analysis(original_ratings):
size_list = []
for user in original_ratings:
size = len(supp_user(user))
size_list.append(size)
size_list.sort()
x = [i for i in range(len(size_list))]
plt.figure()
plt.plot(x, size_list)
plt.savefig('size.jpg')
portion = [0] * 10
for size in size_list:
index = size // 100
portion[index] += 1
logging.info(portion)
logging.info(sum(size_list) / len(size_list))