Title 1.3. Example of Causal Study.pdf ✅

pacmann.io © 2022 – Pacmann AI 21 Example of Causal Studies
pacmann.io © 2022 – Pacmann AI 22 Cash Trip Uber • Given Uber operates in cities where people have fewer credit cards • Problem: ○ Difficulties to make change, drivers may ran out change ○ Drivers paying service fees, the service fees usually charge in credit cards bills, if most transactions are cash, the driver have to pay it forward
pacmann.io © 2022 – Pacmann AI 23 Cash Trip Uber • Drivers don’t like cash trips • Solution: built new feature that tell driver whether a trip is cash when they receive request of trip “Cash” if the request trips is cash, nothing if it was not cash Nick Jones, Sam Barrows: Uber's Synthetic Control | PyData Amsterdam 2019
pacmann.io © 2022 – Pacmann AI 24 Cash Trip Uber Objective: Test whether the new feature affect metric like the unpaid service fees One option where we can test it is by doing A/B Testing It can measures the difference between metric in two different groups Nick Jones, Sam Barrows: Uber's Synthetic Control | PyData Amsterdam 2019 Have New Feature Treatment Group Do not Have New Feature Control Group