def most_frequently_bought_by_user(data_set, user_id):
repository = Repository(data_set=data_set)
user_products = repository.get_products_bought_by_user(user_id)
recommended = sorted(user_products,
key=lambda item: item['count'],
reverse=True)[:20]
for p in recommended:
product = repository.get_product(p['_id'])
if product:
logger.info("{} {}".format(product['product_name'], p['count']))
def analyze_products_by_user(data_set):
repository = Repository(data_set=data_set)
users = repository.get_users()
count = 0
total = len(users)
for user_ids in batch(users, 100):
users_products = []
for user_id in user_ids:
user_products = repository.get_products_bought_by_user(user_id)
user_products = dict(user_id=user_id,
products=[
dict(product_id=p['_id'],
count=p['count'])
for p in user_products
])
users_products.append(user_products)
count += 1
logger.info("{}/{}".format(count, total))
repository.add_user_products(users_products)
