产品市场扩散行为预测的自学习方法

Self-learning approach of product diffusion behavior forecasting

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.