客户分析中的数据挖掘算法比较研究

Technology of Data Mining in Client Analysis

对回归预测、决策树、神经网络、聚类和邻点预测、规则导引等5种数据挖掘预测算法分别进行介绍,并结合实例对各种方法适用情况进行了比较,以便有针对性地对客户行为采用有效的预测方法.其中:回归预测根据历史记录分析得出总体趋势;决策树方法是一种“二分制”数据分析和预测方法,主要用于对数据进行归类分割和预测,来解决定性分析的问题;神经网络方法主要对客户行为进行分析和预测,从定量的角度进行分析.

In this paper, five popular forecasting algorithms of data mining are discussed separately. At the same time, the situations fit for the algorithms are compared combined with the examples. Then, the efficient forecasting methods can be adopted when the different situations of clients are analyzed. Regression forecasting which oftendeduces the general trend according with the historical records is traditional. The Decision Tree method is a data analysis and forecasting method. It is used mainly to divide in classification and forecast so as to solve the problem of qualitative analysis. The method of Neural Networks mainly analyzes and forecasts the clients＇ behaviors with the quantitative point of View.