.png){kind=icon}
.png){kind=icon}
Return
Online procurement problem with risk hedging
.png){kind=icon}
PRE
AI
DOI：10.1016/j.cie.2021.107909.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
We study risk hedging in the online procurement problem, where one buyer and one seller sign reservation contracts in advance to hedge against the risks from the uncertainties of real demand and spot price. We generalize the traditional two-stage procurement model such that the buyer has the flexibility of signing multiple contracts in the first stage and makes decision on executions in the second stage. A lower bound on the performance of any deterministic algorithms is presented by restricting the possible execution price sequence and real demand in worst case. Then a deterministic online algorithm called Price Look Back (PLB) is designed with a competitive ratio that tends to the lower bound if historical execution prices are less fluctuant. Extensive experiments are conducted on both synthetic and real-life instances to evaluate the practical performance of PLB. The experimental results show that PLB performs significantly better than its (worst-case) theoretical guarantee in most instances. In addition, PLB is robust with respect to inaccurate information learned from the historical data. Specifically, the average extra cost incurred by the inaccurate information can be around 5% in real-life instances.
Keywords:
Procurement
Risk management
Demand uncertainty
Operations strategy
Combinatorial optimization
Journal
IF:
6.5
Papers:
1.0W
Citations:
3.8W