def _analyze_users_similarity(args):
user, data_set, min_similarity, offset, limit = args
logger.info("{} {}".format(offset, limit))
repository = Repository(data_set=data_set)
progress = Progress(limit - 1)
for users in batch(
repository.get_users_products(offset=offset, limit=limit - 1),
1000):
for user2 in users:
if user['_id'] == user2['_id']:
continue
progress.advance()
similarity, common, additional1, additional2 = calculate_products_similarity(
user['products'], user2['products'])
if similarity >= min_similarity:
similar = dict(user1_id=user['user_id'],
user2_id=user2['user_id'],
similarity=similarity,
common_products=common,
add_products1=additional1,
add_products2=additional2)
repository.add_users_similarity(similar)
logger.info("{:.1f}% ETA {}".format(progress.get_progress(),
progress.get_estimated_time()))
def analyze_users_similarity(data_set, samples):
repository = Repository(data_set=data_set)
progress = Progress(math.ceil(((samples - 1) * samples) / 2))
offset = 1
for user_products1 in batch(
repository.get_user_products(limit=samples - 1), 100):
for up1 in user_products1:
max_similarity = 0.1
similar = None
for user_products2 in batch(
repository.get_user_products(offset=offset,
limit=samples - offset), 100):
for up2 in user_products2:
progress.advance()
similarity, common, additional1, additional2 = calculate_products_similarity(
up1['products'], up2['products'])
if similarity > max_similarity:
max_similarity = similarity
similar = dict(user1_id=up1['user_id'],
user2_id=up2['user_id'],
similarity=similarity,
common_products=common,
add_products1=additional1,
add_products2=additional2)
logger.info("{} {} {}".format(similar['user1_id'],
similar['user2_id'],
similarity))
if similar is not None:
repository.add_users_similarity(similar)
offset += 1
logger.info("{:.1f}% ETA {}".format(progress.get_progress(),
progress.get_estimated_time()))