You are productizing a batch machine learning pipeline that makes product recommendations. Transaction data from all customer stores is available as an input in the following format:

customer_uuid     date         store_id product_sku  item_count usd_total
34a1a589-cc39-... 2018-01-01.. 18       12421        1          15.25

In addition, product catalog data is also available:

product_sku  product_category
12421-345                Food
10343-124               Latte

Design a batch data processing pipeline that runs nightly and computes features for the ML model to train on. Specifically, for every customer, calculate the following:

1. weekly_food_purchases, weekly_nonfood_purchases: for the past 8 week period (56 days), how many food and non-food items has each customer purchased per week (on average)

2. weekly_spend: for the past 8 week period (56 days), how much money has each customer spend per week (on average)

For a bonus task, compute weekly_visits, how many times has each customer visited the store per week (on average).