摘要
为解决单一扩散模型难以可靠预测产品市场扩散行为的问题,为决策者提供全面的信息和客观的预测结果,提出了一种运用自学习方法确定不同模型的组合权重的方法。给出了对产品市场扩散行为预测的相关描述,运用自学习方法表示了扩散模型库评价向量,确定了组合模型的权重,给出了扩散模型库的自学习方法及具体算法。最后,以移动通信扩散行为预测和互联网扩散行为预测为例,分析了上述方法的有效性。结果表明,上述方法有较好的预测效果。
It was difficult to predict product diffusion behavior reliably by single diffusion model. To provide comprehensive information and objective prediction results for decision-makers, a method to define combinatorial weights for different models based on self-learning approach was proposed. Relevant descriptions for product diffusion behavior forecasting were defined. Evaluation vector of diffusion model base was expressed by self-learning approach. Weights of combinatorial models were also determined. Sell'learning approach and algorithm for diffusion model base was presented. Finally, effectiveness of the above-mentioned methods was proved by diffusion behavior forecasting of Internet and mobile communication. Results demonstrated that the proposed methods could obtain satisfactory forecasting results.
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2008年第12期2484-2491,共8页
Computer Integrated Manufacturing Systems
基金
国家自然科学基金资助项目(60574062)
国家863计划资助项目(2007AA04Z112)
南京农业大学工学院引进人才启动基金资助项目(RCQD07-01)~~
关键词
产品市场扩散
预测
自学习
组合权重
product diffusion
forecasting
self learning
combinatorial weights