基于数据挖掘的电信潜在换机客户的预测研究

发布时间：2019-01-04 06:40

【摘要】：随着数据库技术的广泛应用,各行业积累了海量的业务数据。为了改变当前“数据丰富却知识匮乏”的状况,数据挖掘技术得到了众多企业的重视。在现今移动数据时代、智能终端时代下,智能手机越来越普遍,用户换机的频率和对手机的依赖性更加强烈。所以,在手机终端精准营销方向,对更换手机的潜在客户预测这项研究是有意义的。在这背景下,本文提出基于数据挖掘技术对换机客户进行预测研究的方法。首先,本文依托X省公司“手机终端精准营销需求”项目,分析和采集用户行为数据,并借助Hadoop对真实用户上网日志进行分析。参照CRISP-DM挖掘流程对采集的数据进行数据理解、数据清洗、数据转换等工作。其次,本文分别采用决策树C5.0、神经网络、Logistic回归算法对样本数据进行训练并建立预测模型,并对训练结果进行评估和比较。通过SPSS的实验结果表明决策树C5.0算法模型对潜在换机预测研究更为适合。最后,本文分析了模型对于用户市场拓展、开展终端精准营销、提升业务推荐成功率、用户终端行为监控等方面简单的应用情况。本文从实际问题出发,将数据挖掘技术应用到潜在手机终端更换的预测研究中,研究工作对决策及市场人员开展工作有重要的作用,对同类型换机预测研究有一定参考意义。但本次研究还存在可以改进的地方如评价指标、业务知识,数据处理方法,这些也是将来可开展的研究工作。
[Abstract]:With the wide application of database technology, various industries have accumulated massive business data. In order to change the current situation of "data rich but lack of knowledge", data mining technology has been paid attention to by many enterprises. In the era of mobile data and intelligent terminal, smartphone is becoming more and more popular. So in the precise marketing direction of mobile terminals, it makes sense to predict the potential customers of mobile phone replacement. Under this background, this paper puts forward a method of forecasting customers based on data mining technology. First of all, this paper relies on X province company "the mobile phone terminal precision marketing demand" project, analyzes and collects the user behavior data, and carries on the analysis to the real user online log with the aid of Hadoop. According to the CRISP-DM mining process, data understanding, data cleaning, data conversion and so on. Secondly, the decision tree C5.0, neural network and Logistic regression algorithm are used to train the sample data and establish the prediction model, and the training results are evaluated and compared. The experimental results of SPSS show that the decision tree C5.0 algorithm model is more suitable for the prediction of potential machine change. Finally, this paper analyzes the simple application of the model for user market expansion, terminal precision marketing, promotion of business recommendation success rate, user terminal behavior monitoring and other aspects. Based on the practical problems, this paper applies the data mining technology to the prediction research of the potential mobile phone terminal replacement. The research work plays an important role in the decision making and the work of the market personnel, and has certain reference significance for the prediction research of the same type of machine exchange. However, there are still some areas that can be improved in this study, such as evaluation index, business knowledge, data processing methods, which are also the research work that can be carried out in the future.
【学位授予单位】：云南大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TP311.13

【参考文献】