基于BP神经网络构建的分类器模型财务分析领域的应用  

Application of classifier model based on BP neural network in financial analysis field

作　　者：程书强 罗娟 王念东 

机构地区：[1]陕西科技大学经济与管理学院

出　　处：《自动化与仪器仪表》2018年第1期129-131,134,共4页Automation & Instrumentation

摘　　要：当前，海量数据的集成发展，对数据识别与分析提出了更高的要求，分类挖掘作为数据挖掘技术的重要内容之一，其通过构造分类器来实现数据的深度挖掘与处理，在各领域中均有广泛应用，为此，本文选用精确度高、鲁棒性好的BP神经网络作为分类器模型的构造方法，分析其建模流程，并融合Adaboost算法，针对分类中的FNR和FPR两类错误，引入一种基于重视度的权重自适应算法，以改善其重视度高的样本分类能力，并将其应用至财务预警系统，经Matlab验证加强型的分类器模型具有高效、精准的分类能力。at present, the integrated development of mass data, puts forward higher requirements for the identification and analysis of data, classification mining was one of the important contents of data mining technology, it through constructing the classifier to realize data depth mining and processing, widely used in various fields, due to this,this paper choose BP network which has high accuracy and good robustness as the method to construct the classifier model, analyzed its modeling process, and combined with the Adaboost algorithm, according to the FNR and FPR of two types errors in the classification, introducing a weighted adaptive algorithm based on the degree of attention, in order to improve the sample classification capability of a high degree of attention, and apply it to the financial early warning system, Through Matlab verification, the enhanced classifier model has high efficiency and accurate classification ability.

关 键 词：BP神经网络 数据挖掘 分类器模型 财务分析 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]