Research About Anti-Fraud Detection Of Vehicle Insurance Claims Based On Data Mining Technology

With the spread of the globalization of insurance fraud,the means of insurance fraud are constantly changing,and the identification of insurance fraud is becoming more and more difficult and costly.Motor vehicle insurance fraud accounts for the largest Percentage of insurance fraud.Therefore,based on the literature analysis both at domestic and foreign,the research about Anti-fraud detection of Vehicle Insurance Claims Based on data mining technology has important theoretical significance and application value.Firstly,this paper introduces the current situation of vehicle Insurance Claims fraud,researches background and significance of vehicle insurance claim fraud at domestic and foreign,then analyzes the theory of auto insurance claim and insurance fraud and data mining respectively.According to the characteristics of fraud data samples,three data mining models,Logit model,Bayesian model and decision tree model,are selected.This paper studies the principle of the three model algorithms.Secondly,in the empirical part,the useful fraud factor of Sunshine Insurance Company in 2017 are extracted,through a large number of modeling experiments,using the machine learning method to train and predict the data samples through different data mining models,and verifies the model effectiveness,Finally,this paper gets the conclusion that the most suitable data mining model for the selected vehicle insurance claims data samples is the decision tree C5.0 model with booting method..In the end,expounds the application of data mining technology to anti-fraud detection of vehicle insurance claims in the actual work.and summarizes the suggestions of anti-fraud detection of vehicle insurance claims.