利用交叉覆盖算法对人寿保险客户进行信用评估  

Using Alternative Covering Design Algorithm to Valuate the Reputation of Life Insurance Customer

作　　者：周华[1] 宓浩[1] 

机构地区：[1]安庆师范学院教育学院,安徽安庆246133

出　　处：《安庆师范学院学报（自然科学版）》2009年第2期51-54,共4页Journal of Anqing Teachers College(Natural Science Edition)

摘　　要：在海量的保险数据中进行信息挖掘,可以支持企业决策和降低企业投资风险。神经网络已经广泛应用于这一领域,然而,传统神经网络存在着学习过程中收敛速度慢、网络性能差、可能存在局部极小值等弊端,交叉覆盖算法具有良好分类功能,可对人寿保险客户进行信用评估,对客户的欺诈行为趋势给以预测,结果表明,该方法预测精度很高,且速度快,具有很好的效果。With the huge increment of insurance data mining information, neural network technology can support the enterprise decision--making and reduce investment risk. However, the traditional neural network has some weakness, for example, the slow convergence of learning process, the poor network performance, the local minimum value maybe have, and so on. In this paper, we use the Alternative Covering Design Algorithm, which can not only avoid the weakness a- bove, but also have good function of classification to valuate the reputation of life insurance customer and predict the trend of cheat behavior of customer. The result predicts that the precision of this method is very high, and the speed is quick. It obtains good effect.

关 键 词：神经网络 数据挖掘 人寿保险 覆盖算法 信用评估 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]