Member-only story
Game and Auction Theory for Data Scientists
Part 1 of 6: Payoffs, Best Responses and Nash Equilibrium, how to reason about an outcome that depends on someone else’s decision as well as your own.
Diogo Ribeiro is a mathematician, researcher and lecturer at the Faculdade de Media Artes e Design, Universidade Técnica do Porto. ORCID 0009–0001–2022–7072. GitHub diogoribeiro7. Gitlab diogoribeiro7.
Most of the models a data scientist builds share an assumption that is rarely written down: the environment does not answer back. A churn model does not change the customers it scores. A demand forecast does not alter the weather. The data come from some distribution, the model learns it, and a decision is taken against it.
A good part of applied work does not fit that description. A price is set against competitors who will see it and reprice. A bid is placed in an auction against other bidders who are solving the same problem from the other side. A fraud rule is deployed against people who will probe it until they find what it misses. A credit score is published and applicants change what they report. In each of these, part of the environment is another decision-maker with objectives of their own, and the distribution the model learned was generated by behaviour that will change once the decision is taken.
Game theory is the mathematics of that situation. Auction theory is its most developed application and the…
![[1993] Before the tap: the strange old world when bank cards made an impression](https://webproxy.poorya-velaei-d67.workers.dev/https://miro.medium.com/v2/resize:fit:1358/format:webp/1*GYMcJdV1NyIUi6gPizqttg.jpeg)
