Optimizing ride-hailing fares using computational microeconomy tools
- Prelegent(ci)
- Michał Pawłowski
- Afiliacja
- University of Warsaw
- Termin
- 12 stycznia 2023 12:15
- Pokój
- p. 4050
- Seminarium
- Seminarium „Gry, mechanizmy i sieci społeczne”
Pricing ride-hailing fares sparked a rich line of work in computer science due to its complexity and multidimensionality of aspects. We add to this line of research by considering the model presented by Hikima et al. (AAAI, 2021). For this model, the authors claimed a 3-approximation of the optimum value. Later on, it was improved by Brubach et al. (NIPS, 2022), who got a (1-1/e)-approximation. We take on this problem but approach it from a different angle — we look at it in the context of mechanism design. It allows us to devise a (1-1/e)-approximation algorithm as well. However, we consider our algorithm to be stronger in terms of approximation guarantee than the one of Brubach et al. This is because we compare our algorithm against the optimal Bayesian mechanism, while the previous work was only living in the realm of posted-price mechanisms.