Abstract
Semi-parametric dynamic contextual pricing Page 1 Semi-parametric dynamic contextual pricing Virag Shah — Jose Blanchet — Ramesh Johari Stanford University virag@stanford.edu September 27, 2019 1 / 15 Page 2 Dynamic pricing ▶ Several platforms have access to data describing history of different users. ▶ Platforms can leverage this information to price different products and optimize revenue. ▶ This requires learning online the mapping from user context to optimal price, efficiently. 2 / 15 Page 3 Basic Framework ▶ Discrete times 1,2,...,n, one user arrives per time step ▶ Each user is shown one product, which is ex-ante fixed ▶ Let Vt be the value tth user assigns to the product. ▶ Let pt be the price set by the platform. ▶ The user buys the product if pt ≤ Vt. ▶ Platform does not know or observe Vt, but has access to covariates Xt ∈ Rd which may describe user’s history and product’s type ▶ Goal: set …
Authors
R Johari, J Blanchet
Publication date
2019/12/5
Journal
Advances in Neural Infor
