基于细精度关联规则挖掘的电信客户流失分析

PDF下载HTML

Title:

Analysis of telecom customer churn based on fine-grained association rule mining

作者:

梁路, 王彪, 王剑辉, 刘冬宁

广东工业大学 计算机学院, 广东 广州 510006

Author(s):

LIANG Lu, WANG Biao, WANG Jianhui, LIU Dongning

Faculty of Computer Science, Guangdong University of Technology, Guangzhou 510006, China

关键词:

电信客户流失; 细精度; 关联规则; 逻辑方法; OCAT; 启发式规则

Keywords:

telecom customer churn; fine grain; association rules; logic method; one clause at a time (OCAT); heuristic rules

分类号:

TP182

DOI:

10.3969/j.issn.1673-4785.201404050

文献标志码:

A

摘要:

用决策树等常规关联规则方法分析电信客户流失问题时,存在属性相关性不够精细的问题,即无法剖析属性的内在结构、内涵及隐藏的细粒度的相关规律,同时也无法满足海量电信数据分析的需求.采用细精度关联规则挖掘解决上述问题,从逻辑学角度提出用二进制编码的方法对属性进行分解,用其构造正负训练样本集,然后进行OCAT关联规则挖掘,并加入启发式规则加快收敛速度,以节省时间和内存开销.实验结果表明,基于这种方法产生的关联规则提高了细精度,同时易于实施并行计算和提高效率,能更好地满足当前电信应用需求.

Abstract:

When using traditional association rule mining such as decision tree to analyze the problem of telecom customer churn, we always meet the problem that the dependency of attributes are not enough fine, which means traditional methods not only cannot analyze the internal structure and hidden fine-grained related rules of attributes, but also cannot satisfy the needs of analyzing massive telecom data. In this paper, we solve the above problems by using fine-grained association rule mining. We firstly design a binary coding method from logic viewpoint to break attributes to segments, and then build the positive and negative training sample sets based on segments. In experiment we adopt the one clause at a time (OCAT) algorithm on association rule mining for speeding up the convergence speed and saving the overhead of time and memory. Finally, the experimental result shows that this method improves the fine-grained of the association rule, which can be easily used in parallel computing to raise efficiency, and satisfy the requirements of current telecom application.