基于粗糙集理论的电信客户价值评价研究

发布时间：2018-05-17 20:33

本文选题：电信客户价值评价 + 客户生命周期价值　；参考：《南京邮电大学》2017年硕士论文

【摘要】：随着供给侧结构性改革的推进,面对业务资费的加速下降和用户结构的持续升级,对客户价值的精确描述和评价已成为电信企业的首要任务。然而,面对海量的电信客户数据,传统方法过多依赖领域专家的先验知识,且无法很好的处理数据的不完备、不确定以及评价体系的高维度问题。因此,本文引入改进的粗糙集模型,优化了电信客户价值评价方法。首先,界定了电信客户价值的概念并分析其影响因素,依据客户终生价值思想,创新性的构建了电信客户价值三维评价体系,以实现对客户生命周期价值全方位、多维度的评价;依照指标体系构建的原则,选择具有代表性的客户属性作为各个维度的子指标,并给出具体的计算方式。其次,针对评价体系和电信客户数据的特点,对传统的粗糙集模型进行优化,创新性的将限制优势关系引入加权多粒度粗糙集中,构建了限制优势关系下的加权多粒度粗糙分析方法,并讨论了基于下近似不变的属性约简新算法,同时给出决策规则的获取方式,通过具体实例证明了其优越性;并结合以上分析,梳理了以改进粗糙集模型为基础的电信客户价值评价流程。最后,引入某电信运营商的真实客户数据进行实证分析,初步得出了关键客户属性,并输出相应的决策规则。实践证明,该方法不仅适用于电信客户数据的海量、不完备、高维度等特征,并且实现了基于数据本身的自主化学习,节省了领域专家的重复工作,因此,可以作为电信运营商评价客户价值的决策依据。
[Abstract]:With the development of supply-side structural reform, the accurate description and evaluation of customer value has become the primary task of telecom enterprises in the face of the accelerated decline of service charges and the continuous upgrading of user structure. However, in the face of massive telecom customer data, the traditional methods rely too much on the prior knowledge of domain experts, and can not deal with the incomplete data, uncertainty and high dimensional evaluation system. Therefore, an improved rough set model is introduced to optimize the evaluation method of telecom customer value. First of all, it defines the concept of telecom customer value and analyzes its influencing factors. According to the customer lifetime value thought, it creatively constructs a three-dimensional evaluation system of telecom customer value, in order to realize the all-round value of customer life cycle. According to the principle of index system construction, the representative customer attribute is selected as the sub-index of each dimension, and the concrete calculation method is given. Secondly, according to the characteristics of evaluation system and telecom customer data, the traditional rough set model is optimized, and the restricted advantage relationship is introduced into weighted multi-granularity rough set innovatively. The weighted multi-granularity rough analysis method based on restricted advantage relationship is constructed, and a new attribute reduction algorithm based on lower approximation invariance is discussed. At the same time, the method of obtaining decision rules is given, and its superiority is proved by an example. Combined with the above analysis, the paper combs the telecom customer value evaluation process based on the improved rough set model. Finally, by introducing the real customer data of a telecom operator for empirical analysis, the key customer attributes are preliminarily obtained, and the corresponding decision rules are outputted. The practice shows that this method is not only suitable for the characteristics of telecom customer data, such as magnanimity, imperfection and high dimension, but also realizes autonomous learning based on data itself and saves the repeated work of domain experts. It can be used as a decision basis for telecom operators to evaluate customer value.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F626;F274

【参考文献】