Design And Realization Of Risk Analysis System Of Agriculture-related Loans

With the rapid development of business types and quantity for the ruralcredit cooperatives, the stored data of computer is increasing sharply inrecent years. Nowadays, a large number of valuable historical data is storedin the Science and Technology Centre of the city, however, it has not beenfully mined and utilized. In the current time with data warehouse and datamining technology becoming more developed, how to make good use of thisvaluable resource is a big issue for the rural credit cooperatives. In thecredit assets, the abundant loans issued by the rural credit cooperatives canbe classified into agricultural-related loans, and they majory include thepeasants’ loans, the loans for the rural enterprises and their variousorganizations, and the agricultural-related loans for the urban enterprisesand their various organizations. For these enterprises, the involved personalincome spending is mostly settled in cash, their financial system is not verysound, and there is no financial statements or the authenticity of thefinancial statements is poor. Therefore, the capital flows for the mostenterprises cannot be monitored, and it is hard to recover the loan principaland interest in time when the internal risk happens, which results in highloan risk coefficient.For analysis of single loan or a single loaner, it has many uncertainfactors, big difficulty, strong subjectivity and poor convincing. However,based on a lot of historical data, using data mining technology including theadvanced analysis and modeling technology to summarize and classify the loanrecord and relevant information (e.g.: usage, subordinate industry andregion,etc.), it can find out more abundant information, look for potentiallaw and development trend, and derive the repayment situation for thedifferent classification loan. The above analysis of risk is objective andhas much stronger persuasive. Firstly, it can obtain repayment situationdirectly from the real data regardless of the specific reasons, and avoidsubjective judgment deviation; secondly, it can execute effective collectionbefore the risk occurs for the similar loans by observing their trend; thirdly,it can supply a certain reference to determine the release of new loans ornot, credit limit and time based on adverse proportion in the differentclassification loans.This system applies DB2v9database product of IBM company, establishesrelated database and table according to the analysis results of system demand, imports all of the current related historical data initially by writing shellprogram, later changes increment and inserts related data every day. Thesystem uses DWE Intelligent Miner data mining analysis tools for thepreliminary mining processing, and lets the business staff to give reasonableexplanation to the mining results according to the actual situation andeliminate the abnormal and garbage information, consequently to ensure themining results are true and effective; Meanwhile, it employs On-LineAnalytical Processing (OLAP) technology for the argument, and uses DWE CubeViews to conduct data mining operations (e.g.: roll-up and drill-down) byestablishing multidimensional data cube model for the mining results fromartificial analysis and processing. By writing relevant shell in the databaseserver, it realizes the data collection and pretreatment; moreover, applyingEOS developing and integrating environment of Shanghai Primeton company todevelop application program, it achieves that the front desk users canaccording to authority freely choose the bank branch and concept application(loan application) to inquires classification of loans in the specific dateor time period. It provides services with web form in the internal network,so that it is convenient to inquire and use for the non-technical staff suchas decision makers and ordinary employees.The purpose of developing this system is to help rural credit cooperativesto strengthen risk management of credit assets, get valuable informationthrough the analysis of the historical data, summarize and analyze the loansbased on different classification, monitor the stock loans according to therepayment proportion trend, execute loan collection in advance, determine therelease of new loans or not and loan limit according to the repayment trend.In theory it can obtain repayment trend of different classification loansthrough the analysis of historical data, effectively monitor the stock loans,reduce production of bad loans, add real and effective evidence to the newloan approval, and effectively lower the risk coefficient of theagriculture-related loans in practice. This study makes a certaincontribution to improve theoretical level of agricultural-related loans,promote the regular development of the business and the healthy growth of thecustomer managers for rural credit cooperatives.