The outlet team needs a notebook they can read without translating between Python objects and Spark columns. Work with its separate table, main.silver.outlet_orders. Each answer should display the final DataFrame, not print the text of a SQL query.
You will combine the transformations from the previous lessons: select columns, filter records, rename a field, normalize text and handle missing values. The same operation can be useful or harmful depending on the prompt. Filling an unknown amount is right for a zero-default report and wrong for a completeness audit that must preserve NULLs. Read that distinction before writing the expression. The outlet extract also has repeated customers, so a list of buyer keys needs deduplication even when the order ids are unique.