神经网络算法是数据挖掘的核心技术之一,在金融业中有着非常广泛的应用前景。本文提出了一种改进的BP网络算法,并将该算法应用到邮政账单分类中,通过分析用户集合构成规格化的数据库,获得满足最小支持度的项集,进而形成规则集。利用邮政账单数据进行测试,得到了精确的具有相同特性的客户组,实验表明该方法是行之有效的。 Neural network algorithm is one of the core technologies of data mining. It has a very wide range of application prospects in the financial industry. In this paper, an improved BP network algorithm is proposed. The algorithm is applied to the postal bill classification. By analyzing the user set to form a normalized database, the item set satisfying the minimum support is obtained, and then a rule set is formed. Using the postal bill data for testing, we get the exact customer groups with the same characteristics. Experiments show that this method is effective.