| Based on the study of noise information effect on credit risk of enterprises, this paperdiscusses the impact of credit risk on enterprise. By building the concepts and quantitativestructure of credit risk distortion and considering the "disclosure the false information", westudy the early warning of credit risk in connection with the moral factors of enterprise"fraudulent sales".The master’s thesis research is part of the "With noise interference information distortionof enterprise credit risk study on the construction and calibration strategies"(2011-2013,Grant No.10YJC630334)" of the Ministry of Education Humanities and Social SciencesProject. The following issues are mainly studied and addressed in this thesis based onanalyzes the research of enterpris credit risk:1. Define the distortion and the degree of distortion of enterprise credit risk.2. The paper derives the neural network model of time series of multi-sample andmultidimensional indexes by introducing the double factors of enterprise itself and similarindustry on the research of enterprise credit risk distortion, furthermore, designs a dimensionreduction method for the information data based on Hodrick-Prescott (HP) filter fordiscriminating the credit risk distortion.3. In order to consider the impact of false informationon to the enterprise credit riskfurther, the paper derives the measure structure of credit risk distortion by extracting the ideaof outlier mining and vector space model, and then gives an empirical experiment in whichthe results show the validity of the measure structure.4. This paper considers the early warning of enterprise credit risk with the moralfactor,building the concept of the enterprise moral credibility to quantify the level ofenterprise moral with the enterprise fraud probability of selling,and applies the result to builtthe Logistic regression model in early warning of credit risk.Finally,summarizes the main results of this study and proposed the direction for furtherresearch. |
PDF Full Text Request