供应链线上企业信贷风险动态聚类投影寻踪建模与实证研究

Dynamic Clustering Projection Pursuit Modeling for Credit Risk Assessment of the Corporates on Supply Chain and Its Positive Research

根据供应链线上核心企业情况等六个一级指标、信用记录等17个二级指标和核心企业净资产收益率等25个三级指标构成的评价指标体系和采集到的样本数据，应用动态聚类投影寻踪（DCPP）模型对12家基于供应链的企业信贷风险进行实证评估研究，建模结果表明：DCPP模型结果与低维逐次投影寻踪模型的结果完全一致，与专家法结果的一致性好，且不受人为主观因素的影响，又能同时求得企业信贷风险和评价指标权重的大小及其排序，能很好地应用于供应链线上企业信贷风险评估研究，具有可操作性和客观性好以及后续应用简捷等特点，是一种企业信贷风险有效和精细评估的新方法．

Based on the six credit risk assessment aspects such as the core enterprises sit- uations et al, the seventeen second indexes such as the credit records et al, the twenty-five assessment indexes such as the core enterprise return on equity （ROE） et al, as well as the sample data of the twelve enterprises, the dynamic clustering projection pursuit （DCPP）model for the credit risk assessment of the enterprises on the supply chain is established in this paper. The case study shows that the DCPP model is well suitable to the credit risk assessment of the enterprises on the supply chain, and the results of the DCPP model is good agreement with that of the low-dimensional sequential PPC model, as well as that of the expert method. The magnitude and the rank of the credit risk of the enterprises and of the assessment indexes are simultaneously obtained by optimizing the DCPP model. The DCPP model possesses good objective characteristics and practices, and being convenience in the following usage. The DCPP model is a new methodology to assess the credit risk of the enterprises effectively and subtly.