摘要
以客户关系管理(custom er re lationsh ip m anage-m en t,CRM)的数学模型为背景,研究了如何用Hop fie ld神经网络构建一类M arkov链表述的CRM的客户分类分析和评价的计算模型。从一种不同于现CRM的对客户关系评价的思路入手,用M arkov链建模。分析该类马氏链建立的CRM数学模型的特点(无限次交易),分析连续Hop fie ld神经网络计算的内在特点。从矩阵结构和求逆的角度,发现这两个不同概念模型的数学模型具有相同的特点。研究结果将该类马氏链的CRM模型计算同Hop fie ld神经网络计算关联起来。这表明可以用连续Hop fie ld神经网络计算该类M arkov链的CRM模型。
A Markov chain model was used to describe a customer relationship management model using a continuous Hopfield neural network model to analyze customer classifications and to evaluate plant sales. The Markov chain model gives very different evaluations than the origin customer relationship management model. The characteristics o{ the management model (in the infinite horizon case) are mathematically similar to those of the neural network model. This study combines the Markov chain model with the neural network model to show that the neural network can be used to analyze the Markov chain model.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2005年第12期1696-1699,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金重点资助项目(70231010)