The cafe's new till exported cafe_export_raw.csv, loaded as raw. It has every problem from this unit: item names in random capitals with stray spaces, payment methods spelled four ways, prices stored as text with a dollar sign, some missing quantities and prices, and dates written in two formats (2026-04-06 and 04/06/2026).
That last one is new: pd.to_datetime reads the first value to guess the format, then fails on rows written differently. format="mixed" tells it to work each value out separately.
Each exercise returns a cleaned column or an answer; you can combine the steps in later cells as you like.