摘要
研究电子商务零售中动态捆绑决策的优化问题.延迟购买行为会大大降低易逝品零售商的收益水平,动态捆绑策略是有效解决延迟购买效应的一种营销手段,计算复杂性和决策的实时性需求是实施动态捆绑策略的难点.对此,首先基于延迟购买效用分析提出了易逝品市场需求状态模型和动态捆绑的决策过程模型,并给出了决策过程模型中随机参数的估计方法.在此基础上,根据问题特点对传统Q学习算法加以改进,使之适应动态捆绑策略的优化问题.模拟实验结果表明:1)延迟购买程度越大,动态捆绑策略对收益的贡献也越显著;2)用所提出的改进型Q学习算法求解动态捆绑策略优化问题具有较高的效率和效用.
This paper researched optimization problem of dynamic bundling decision in e-retailing setting. Delay buying behaviors greatly reduce perishable products retailer's revenue level.Dynamic bundling strategy is a marketing tool which can effectively overcome delay buying effect.However,the nature of high computational complexity and the requirement of decision in real time constitute main difficulties for implementing this strategy.For this point,this paper presented perishable products demand state model and dynamic bundling decision process model based on delay buying utility analysis,and then gave the stochastic parameters estimation method for the decision process model.Based on those,this paper improved the traditional Q-learning algorithm according to the problem characteristics so as to make it fit for dynamic bundling optimization problem.The simulation experiment result validates that:(1) the higher extent of delay buying,the bigger revenue contribution can dynamic bundling strategy make;(2) the approach proposed in this paper owns high effectiveness and efficiency for solving the dynamic bundling optimization problem.
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2011年第10期1892-1902,共11页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(70871039)