财务困境预警的二阶段模型构建与应用

Construction and Application of Two-Stage Model for Financial Distress Early Waring

模糊神经网络汇集神经网络和模糊逻辑的优点,能有效避免神经网络的"黑箱"操作,但存在"维数爆炸"现象。将粗糙集和模糊神经网络有机集成,构建财务困境预警的二阶段模型:第一阶段利用粗糙集知识约简对数据集降维消冗,提取最优指标集;第二阶段以最优指标集设计基于模糊神经网络的财务困境预警模型。该模型融合粗糙集和模糊神经网络的特点,能提高网络结构的精练性、启发性和透明性。应用实例的结果表明该模型能有效克服"维数灾难",避免数据噪声引起的模型过度适应,提高模型预测准确性。

Fuzzy neural network which brings together the advantages of neural network and fuzzy logic, can avoid the ＂black-box＂ operation but ex- ist the curse of dimensionality. Proposes a two-stage model for financial distress early warning by integrated the rough set theory into fuzzy neural network. In the first stage, extracts an optimal index set by using the knowledge reduction of rough set to reduce dimension- ality and eliminate redundant of the data set; designs a fuzzy neural network model for financial distress early warning based on the opti- mal index set which is extracted in the first stage. The new model can improve the conciseness, heuristics and transparency of network structure by combined rough set into fuzzy neural network. The application result indicates that the model, which can overcome the ＂curse of dimeusionality＂ and avoid the over fitting that caused by data noise, can obtain much higher accuracy in financial distress prediction.