企业财务危机预警指标选择的质量,直接影响预警结果的准确性。本文采用财务数据结构下层次化定性分析方法,对预警指标信息冗余进行约减,并与采用定量信息约减方法——粗集(Roughset)方法进行对比。以56家商业上市公司为样本,通过采用ANN预警模型进行实证,结果表明了层次化定性分析方法的有效性,从而有助于提高企业财务危机预警模型的准确度与应用性。 The quality of early warning indicator selection of corporate financial crisis has a direct impact on the accuracy of early warning results. In this paper, the qualitative analysis method of financial data structure is used to reduce the information redundancy of the early warning index, and compared with Roughset method, which uses the quantitative information reduction method. Taking 56 commercial listed companies as samples, this paper uses the ANN pre-warning model to conduct empirical research, which shows the effectiveness of the hierarchical qualitative analysis method, which helps to improve the accuracy and applicability of the early warning model of financial crisis.