基于遗传算法的模糊聚类在用户行为分析中的应用研究

发布时间：2018-03-16 06:24

本文选题：数据挖掘　切入点：FCM聚类算法　出处：《山东师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：电信用户行为分析,是指在获得海量电信用户行为数据的情况下,利用数据挖掘等相关技术对电信用户行为进行有针对性的科学分析,从而发现用户消费的一些规律,并将这些规律与电信企业的科学化决策、针对化营销和交叉销售方案的设计相结合。以提高客户黏性需求,更好地服务客户。模糊C-均值(FCM)聚类分析作为数据挖掘中的一项重要数据处理技术,目前被广泛的研究并应用到各大领域中。FCM算法的特点是易于理解,描述简明,实用性强、收敛速度快及自动分类等优点。但由于FCM聚类算法在分析数据时容易陷入局部最小值,为了解决FCM算法这一问题,将遗传算法应用到FCM聚类算法中,因为遗传算法具有全局寻优的能力,且遗传算法易与其他算法结合。所以将遗传算法与FCM聚类分析相结合,依托FCM强大的收敛速度和遗传算法优秀的全局寻优能力,可以很好地应用到电信用户行为分析当中。本文在深入研究FCM聚类算法和遗传算法的基础上,将传统意义上的遗传算法与FCM聚类算法进行了一定程度的优化。在遗传算法优化中主要对遗传算法的适应度函数、遗传算子操作(选择、交叉和变异)及遗传参数的选择进行了一定的优化;在FCM聚类算法优化中主要对相关参数(模糊因子)进行了优化。将优化后的遗传算法与FCM聚类分析相结合,应用到电信用户行为分析当中,本文所涉及到的电信用户行为主要是电信用户的消费行为(通话消费行为、短信消费行为和流量消费行为),通过深入分析这些电信用户行为,进而为合理套餐的提出提供一定参考,并为科学化决策、针对化营销和交叉销售方案的设计提供依据。并用实验分析来分析算法的可行性。
[Abstract]:Telecom user behavior analysis refers to the use of data mining and other related technologies to conduct targeted scientific analysis of telecom user behavior in the case of obtaining massive telecom user behavior data, so as to find some laws of consumer consumption. These laws are combined with the scientific decision of telecom enterprises, aiming at the design of marketing and cross-selling schemes. As an important data processing technology in data mining, fuzzy C- mean FCM clustering analysis is widely studied and applied to various fields. The characteristics of. FCM algorithm are easy to understand, concise description and strong practicability. In order to solve the problem of FCM algorithm, genetic algorithm is applied to FCM clustering algorithm. Because genetic algorithm has the ability of global optimization, and genetic algorithm is easy to combine with other algorithms. Therefore, the combination of genetic algorithm and FCM clustering analysis, relying on the strong convergence speed of FCM and the excellent global optimization ability of genetic algorithm, It can be applied to telecom user behavior analysis. Based on the in-depth study of FCM clustering algorithm and genetic algorithm, The traditional genetic algorithm and FCM clustering algorithm are optimized to a certain extent. In genetic algorithm optimization, the fitness function of genetic algorithm, genetic operator operation, The selection of crossover and mutation) and genetic parameters are optimized. In the optimization of FCM clustering algorithm, the related parameters (fuzzy factors) are optimized. The optimized genetic algorithm is combined with FCM clustering analysis. Applied to the analysis of telecom user behavior, the telecom user behavior in this paper is mainly the telecommunication user's consumer behavior (call consumption behavior, short message consumption behavior and flow consumption behavior), through in-depth analysis of these telecom user behavior, Then it provides some reference for the proposal of reasonable package, and provides the basis for scientific decision, aiming at the design of marketing and cross-selling scheme, and analyzes the feasibility of the algorithm with experimental analysis.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;TP18

【参考文献】