摘要
针对当前大部分企业危机预警模型特征选取和等级分类困难的现状,提出一种基于熵的预警方法.该方法首先对企业运营的指标进行特征选择,在选定指标的基础上利用判别熵选取企业危机预警特征值,然后利用熵聚类算法对预测结果进行分类,判断企业的危机状态,最后采集沪深A股上市公司的年报数据对该模型进行了实证分析,结果表明该模型有效可行.
Due to most of current early-warning difficulties of enterprise crisis in feature extraction and degree classification , the paper presents a method based on entropy. Firstly, J-divergence entropy is applied to feature selection and feature extraction. Then the calculating result is classified to judge state of enterprise with the entropy clustering model. Finally, the results show the feasibility and validity of the model using data acquired from annual reports of A-stock market as samples.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2009年第4期43-49,共7页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(70372011)
关键词
危机预警
特征提取
熵聚类
crisis early-warning
feature extraction
entropy clustering
