Research On The Application Of Data Mining Technology In Enterprise Risk Audit

The development of information technology makes the data-based audit model come into being.Based on this,the core method of audit has changed into data analysis.But in the era of big data,traditional computer-aided audit techniques have been difficult to meet the needs of audit data analysis and we require more powerful data analysis techniques to address this challenge.Data mining can discover valuable information from massive amounts of data and translate it into organized knowledge.In this context,domestic and foreign scholars have been concerned about the subject that how to use data mining technology to better assist the audit.After several years of exploration,the scope of research continues to expand and the extent of research continues to deepen,but it is still in the development and there are still many blank areas in the field.This paper studies the application of data mining technology in insurance company risk audit,and aims at the segmentation of agricultural insurance.Agricultural insurance is a risk dispersion mechanism for agriculture.However,we began to establish the agricultural insurance system from the beginning of 2007.It starts fairly late in our country.Therefore,information asymmetry also exists in the transactions of agriculture insurance which can leads to adverse selection and moral hazard.Especially the information asymmetry of the insured farmers,which makes the insurance companies have high operating cost and high operational risk.Therefore,the insurance companies need to strengthen the audit of insured farmers.However,in the context of information age,a huge agricultural insurance business data are produced every day and the audit has lagged behind the business,so that auditors are unable to identify customer risks timely.At this point,we need to use computer technology to assist the audit.We can give full play to the advantages of easy access to data preservation and use the data mining techniques to find hidden knowledge behind the massive data.As a result,the insurance companies will implement customer risk audit better and response to customer risk more effectively.The goal of this paper is to find an effective solution to the problem of information asymmetry in the agricultural insurance market and the problems faced by the insurance companies in the implementation of customer risk audit.After research,we believe that insurance companies can use the clustering and classification method to construct the customer risk classification model and determine the difference insurance rate according to the division result to solve the above problem.Firstly,this paper introduces the clustering and classification,and reviews the practical application of data mining in risk auditing.Then we focus on the research object of this paper,which is agricultural insurance.We describe the current situation and existing problems of agricultural insurance in China and put forward the corresponding countermeasures,that is,the use of data mining technology to assist customer risk audit.After that,we use R language to carry out data mining for the agricultural insurance data of X City.We construct the customer risk classification model by using of the k-means algorithm and the random forest.The k-means algorithm is unsupervised method and the random forest is supervised method.The paper first describes the theoretical principles,and then introduces the actual operation process,including data preparation,data preprocessing,model building,model utilization and assessment of the model.Based on the clustering model,we determine the risk level of the insured farmers as five categories.Then the model is optimized based on the classification method,so that it can be directly used by auditors and can predict the risk of new insured farmers.After that,we tested the accuracy of the model,the accuracy rate of up to 99.6%.Thus,the customer risk classification model for the agricultural insurance is feasible,which can help insurance company to improve customer risk audit efficiency and avoid the risk of agricultural insurance.Finally,according to the whole research process,we propose some relevant recommendations about the application of data mining technology in customer risk audit.The significance of this paper lies in two aspects.In theory,we compare the preconditions and the effect of clustering and classification,and make full use of the data of the agricultural insurance companies to make the first quantitative study on the risk of the insured.In practice,on one hand,we construct the risk classification model for the X City agricultural insurance company by using the data mining techniques.On the other hand,this paper demonstrate the whole process of data mining clearly.The above can provide a reference for the insurance company to carry out customer risk audit.So that,they can conduct more effective customer risk management and risk control.