Dynamic pricing with finite price sets: a non-parametric approach

Abstract We study price optimization of perishable inventory over multiple, consecutive selling seasons in the presence demand uncertainty. Each season consists a finite number discrete time periods, and per period is Bernoulli distributed with price-dependent parameter. The set feasible prices finite, expected corresponding to each unknown seller, whose objective maximize cumulative revenue. propose an algorithm that estimates parameters learning phase, subsequent applies policy determined as solution sample dynamic program, which modifies underlying program by replacing estimate. Revenue performance measured regret: revenue loss relative optimal attainable under full information. For given n , we show if allocated asymptotic $$(n^2\log n)^{1/3}$$ ( n 2 log ) 1 / 3 then regret same order, uniformly all parameters. An extensive numerical compares our six benchmarks adapted from literature demonstrates effectiveness approach.