主管：中国科学院
主办：中国优选法统筹法与经济数学研究会
　　　中国科学院科技战略咨询研究院
中国管理科学

中国管理科学 ›› 2017, Vol. 25 ›› Issue (11): 66-75.doi: 10.16381/j.cnki.issn1003-207x.2017.11.007

• 论文 • 上一篇    下一篇

基于后悔理论与多Agent模拟的新产品扩散消费者决策互动行为研究

危小超1, 李岩峰2, 聂规划1, 陈冬林1   

  1. 1. 武汉理工大学经济学院, 湖北 武汉 430070;
    2. 同济大学经济与管理学院, 上海 200090
  • 收稿日期:2015-03-12 修回日期:2016-12-30 出版日期:2017-11-20 发布日期:2018-01-31
  • 通讯作者: 危小超(1986-),男(汉族),河南郑州人,武汉理工大学经济学院,副教授,博士后,研究方向:管理系统模拟,E-mail:weixiaochaowin@163.com E-mail:weixiaochaowin@163.com
  • 基金资助:

    国家自然科学基金资助项目（71601151，71172043，71701075）；中国博士后基金面上资助项目（2014M552102）；湖北省自然科学基金资助项目（2014-IV-114）；教育部人文社科基金青年项目资助项目（16YJC630131）

Research on Consumer Decision-making Interaction Behavior in New Product Diffusion Based on Regret Theory and Multi-Agent Simulation

WEI Xiao-chao1, LI Yan-feng2, NIE Gui-hua1, CHEN Dong-lin1   

  1. 1. School of Economics, Wuhan University of Technology, Wuhan 430070, China;
    2. School of Economic and Management,Tongji University,Shanghai 200092,China
  • Received:2015-03-12 Revised:2016-12-30 Online:2017-11-20 Published:2018-01-31

PDF (PC)

626

被引次数

1

摘要： 基于后悔理论，对新产品扩散消费者决策互动行为进行多智能体模拟研究。从进化博弈视角构建了消费者决策行为模型，结合心理学与后悔理论，设计出一种综合考虑自身与邻居个性特征以及历史信息的学习规则，并用多智能体方法对消费者群体决策场景进行仿真。在仿真软件Anylogic6.5.0上进行多Agent模拟实验，模拟结果表明：（1）产品收益b越大，越有利于群体的决策状态稳定，反之，决策波动大；（2）消费者群体决策的宏观趋势由产品扩散成本和因未购买而损失的收益决定。当产品扩散成本c与因未购买而损失的收益f相当时，群体决策行为呈针锋相对态；当扩散成本较高时，拒绝占优；反之，则购买占优；（3）消费者群体间沟通学习强度影响决策波动性，强度高时，群体决策变化大，而低强度下，群体行为稳定。该研究可为新产品扩散营销策略制定提供支持，并为将心理学相关理论与演化博弈应用于Agent模型提出一个基础框架。

关键词: 后悔理论, 进化博弈, 消费者决策, 多Agent模拟, 互动行为

Abstract: The dynamics of new products diffusion not only emerge partly from the interaction among heterogeneous consumers in social ecommerce, but also depend on consumers' psychology behavior. Accordingly, the psychology theory is worth being introduced to explore the evolution of new product diffusion.An agent-based simulation model is built to study the new product diffusion, based on the regret theory and evolutionary game theory, which can be used to support dynamic management for new product marketing.Firstly, an evolutionary game model is proposed to depict the interaction between consumers in social ecommerce, by extending a traditional symmetric game to a dynamic game. Then, the evolution learning rules are designed based on the regret theory, taking the historical information and neighbor characters into consideration. At last, using Anylogic 6.5.0, a multi-agent simulation system is developed, and then a series of simulation experiments are performed and analysed. Simulation results reveal that:(1) The product benefit has a positive effect on the stability of diffusion dynamics.Higher product benefit can result in more stable (volatile) in the diffusion process. (2) The new product diffusion also depends on the cost of product diffusion and the loss from adopting refuse strategy. The group will more likely have the Tit-for-Tat (retaliation) issue when the two variables are equivalent. If the diffusion cost is higher,the dominate strategy is refuse, and vice versa. (3)The communication strength of consumers in market is positively associated with the volatility of diffusion dynamics. The greater communication strength of consumers indicates more obvious volatility of diffusion, while the weaker communication strength more likely result in the stability of group behavior. This research can make some contributions to new product marketing, and provides a new framework to combine psychology theories and simulation models.

Key words: regret theory, evolutionary game, consumer behavior, multi-agent simulation, interaction

中图分类号: