摘要
建立K近邻法案例推理模型的关键在于属性权重的确定。本文使用遗传算法、Lo-gistic回归标准化系数法来确定属性权重,这样就可以避免使用层次分析法、德尔菲法确定权重带来的成本过高的弊端。本文使用我国上市公司376个样本进行财务危机预测分析,预测结果表明,案例推理模型预测准确率高于Logistic回归模型。
The key to establish the case-based reasoning model of K-nearest neighbor is to determine the attribute weights. In this paper we use genetic algorithm and logistic regression coefficient of standardization to determine the attribute weights, so that we can avoid the big cost of using Analytic Hierarchy Process and Delphi method. We established a case-based reasoning model of financial crisis by using 376 samples of Chinese Listed Companies. Compared this model with the logistic regression model, it can identify financial crisis more accurately than the logistic regression model.
出处
《经济管理》
CSSCI
北大核心
2009年第6期123-131,共9页
Business and Management Journal ( BMJ )
基金
教育部人文社会科学一般项目"产业安全监测预警与对策研究"(08JA790053)
吉林大学"985工程"经济分析与预测哲学社会科学创新基地项目"我国上市公司财务危机预测模型研究-基于统计和人工智能方法构建"
关键词
财务危机
案例推理
K近邻
遗传算法
LOGISTIC回归
financial crisis
case-based reasoning
K-nearest neighbor
genetic algorithm
logistic regression