企业担保风险评估方法研究与仿真

摘要

Research accurate risk assessment of enterprise guarantee. The paper put forward a neighborhood rough set model attribute reduction algorithm based on the attribute distribution set neighborhood threshold, which con-structes a fast attribute reduction method and constraines the conflict attribute to reduce errors. Simulation results show that the improved neighborhood rough intensive can obviously improve the accuracy evaluation and use less index variables for getting better evaluating results.%研究企业担保的准确风险评估问题.针对不同担保企业中,受到政策、市场、地域分布等多方面因素的影响,待担保企业的利润量属性、成本属性以及众多属性之间存在较大冲突,传统的风险评估模型多是利用利润量和成本量这两个单项属性完成担保企业估计的,两者的冲突会导致模型稳定性不高,造成担保风险评估准确性降低.为此提出一种邻域粗糙集模型属性约简的算法,根据属性分布特征设定邻域阈值,构建一种快速的属性约简方法,约束冲突属性,减少误差.仿真结果表明,改进后的邻域粗糙集约简明显提高了评估准确性,能用较少的指标变量得到更好的评估结果.