INNER CODE UNIT · Python
main
alteryx/open_source_demos · predict-next-purchase/process_data.py:23
def main():
data_dir = "data"
order_products = pd.concat([pd.read_csv(os.path.join(data_dir,"order_products__prior.csv")),
pd.read_csv(os.path.join(data_dir, "order_products__train.csv"))])
orders = pd.read_csv(os.path.join(data_dir, "orders.csv"))
departments = pd.read_csv(os.path.join(data_dir, "departments.csv"))
products = pd.read_csv(os.path.join(data_dir, "products.csv"))
users_unique = orders["user_id"].unique()
chunksize = 1000
part_num = 0
partition_dir = "partitioned_data"
try:
os.mkdir(partition_dir)
except:
pass
for i in tqdm(range(0, len(users_unique), chunksize)):
users_keep = users_unique[i: i+chunksize]