Research On The Risk Prediction Of Accounts Receivable Based On Bayesian Network

In the increasingly fierce market competition,it’s difficult for business to survive by merely improving product and service quality, reducing production costs, enhancing advertising,etc. Companies have to take credit, installment credit and other transactions in order to expand sales, increase market share, strengthen competitive position. Using credit transactions on the one hand to expand the company’s market share, but on the other hand the formation of receivables. Due to the factors of customer, market and many other aspects of existence, the value of accounts receivable is uncertain, and the accounts receivable is easy to form a bad and doubtful debt,that is the risks of the accounts receivable. Once the bad and doubtful too mach, would seriously lead the capital stand break, endanger the development of the enterprise, and even lead to bankruptcy. Therefore, enterprises in the positive use of credit this marketing means to expand market share, they must strengthen risk control. The base of the risk control is the accounts receivable risk identification and prediction.There are few research in the domestic on the prediction of the accounts receivable risk,and among the research, most take the subjective qualitative analysis method which is not conductive to the precision of the risk prediction. The accounts receivable risk is random event with its uncertainty. With the research starting lately, the account receivable risk’s historical data are quite few. Therefore, we take the Bayesian net work model to predict the account receivable risk. On the base of,Bayesian network can make full use of historical data and expert knowledge which are combined, and it can handle incomplete data, analysis uncertain events and then solve the problem of historical data less.On the base of large literature reading, the definition of accounts receivable risk, the mechanism, as well as the impact on the business and other risk factors are combed in this paper. The related theories of the Bayesian network and principal component analysis used here are introduced. Based on the analysis of the accounts receivable risk factors, considering the accuracy and complexity of the model,we select the key factors and construct a Bayesian network model to predict the risk of accounts receivable. The key factors selected include repayment ability, the way of accounts receivable security,the frequency contacts and the dependence between the buyer and the seller. To simplify the model while ensuring accuracy, the principal component analysis is used to obtain the state of the repayment. Based on the financial data of listed companies, the enterprise repayment ability evaluation model is established. Principal component analysis method can effectively simplify the complexity of the model by reducing the dimensions, and it can effectively avoid the subjectivity of expert opinions method, by using the variance explained rate to determine the index weight. In the end, an example is given to show the process of the model.The accounts receivable risk prediction model is constructed based on the Bayesian network and principal component analysis. The model has many advantages in the use of daily life. The date needed is easy to get, and the model is simple to use. The raw data into the model can directly obtain the quantitative data, and the fuzzy of the calculation results is effectively reduced. The model can be used before the accounts receivable occur to predict the risk, as the reference for the enterprise decision.