def _analyze_orders_similarity(args):
order, data_set, min_similarity, offset, limit = args
logger.info("{} {} {}".format(order['_id'], offset, limit))
repository = Repository(data_set=data_set)
progress = Progress(limit - 1)
for orders2 in batch(repository.get_orders(offset=offset, limit=limit - 1),
1000):
for o2 in orders2:
if o2['_id'] == order['_id']:
continue
progress.advance()
similarity, common, additional1, additional2 = calculate_products_similarity(
order['products'], o2['products'])
if similarity >= min_similarity:
similar = dict(order1_id=order['order_id'],
user1_id=order['user_id'],
order2_id=o2['order_id'],
user2_id=o2['user_id'],
similarity=similarity,
common_products=common,
add_products1=additional1,
add_products2=additional2)
repository.add_orders_similarity(similar)
logger.info("{:.1f}% ETA {}".format(progress.get_progress(),
progress.get_estimated_time()))
def analyze_orders_similarity(data_set, samples):
repository = Repository(data_set=data_set)
progress = Progress(math.ceil(((samples - 1) * samples) / 2))
similarity_threshold = 0.2
offset = 1
for orders1 in batch(repository.get_orders(limit=samples - 1), 100):
for o1 in orders1:
max_similarity = similarity_threshold
similar = None
count = 0
for orders2 in batch(
repository.get_orders(offset=offset,
limit=samples - offset), 100):
for o2 in orders2:
progress.advance()
similarity, common, additional1, additional2 = calculate_products_similarity(
o1['products'], o2['products'])
if similarity > max_similarity:
max_similarity = similarity
similar = dict(order1_id=o1['order_id'],
user1_id=o1['user_id'],
order2_id=o2['order_id'],
user2_id=o2['user_id'],
similarity=similarity,
common_products=common,
add_products1=additional1,
add_products2=additional2)
logger.info("Similarity {} {} {}".format(
similar['user1_id'], similar['user2_id'],
similarity))
if similar is not None:
repository.add_orders_similarity(similar)
offset += 1
logger.info("{:.1f}% ETA {}".format(progress.get_progress(),
progress.get_estimated_time()))
