Research On The Credit Risk Measurement Of Commercial Banks Based On The Data Mining With Clementine

Credit risk is one of the main risks of the commercial banks’ risk. McKinney’sresearch shows that the credit risk is more than three times than the sum of marketrisk and operational risk. As one of the most important factors of the bankingmanagement, it is always a hot research issue during the government and bankresearcher. With the enlargement of the impact of the global financial crisis and credittransactions, commercial banks face more emergency situation about the credit riskwhich is more diversified and complicated. In China, the credit risk measurementaccuracy and sensitivity are both the weaknesses in constraining banks’ risk. How todefine credit risk and how to impose effective risk management measures are not onlythe key problem in commercial banks profit earning but also play important part inprotecting the interests of investors and building harmonious society. In this paper,from the commercial bank’s perspective, constructed a credit risk measure indexsystem, proposed the Credit Risk Measurement of Commercial Banks based on theData Mining with Clementine, in order to provide technical tools and referenceapproach in the Credit risk measurement.Firstly, discussed the theory of measure of credit risk in commercial bank,combined with current conditions to redefine the credit risk of the corporate clients ofcommercial banks from Basel which approved risk management in commercial banks;Introduced the accuracy credit risk model and studied the data mining technology inthe application of credit risk management in commercial banks, especially theadvantages and meanings.Secondly, introduced the processes of the accuracy measurement model on thebasis of the construction of index system. Analyzed the influencing factors, includingfinancial factors, industry, property and macroeconomic of China, obtained40indicator and established a logistic regression model with reverse stepwise selectionscreening variable reduction factor of indicators data analysis.Finally, researched the sample of120selected empirical corporation’s relevantfinancial and industrial data with Clementine which is one of the most popular datamining software. Established a logistic regression model with the KMO and Bartletttest in factor analysis to reduce indicators and according to the "CRISP-DM" data mining principle. Got better accuracy and stability in the random sample of theprediction of the measurement of credit risk with the model. The results showed thatthe data mining technology have better evaluation in predicting the probability in thecommercial bank’s credit risk measurement.