providing a framework for reforming premium rates of vehicle collision coverage with using neural networks model (a case study of asia insurance company)

since vehicle collision coverage, unlike what it seems, is not very profitable for insurance companies and is moving towards making losses, this paper considered the adequacy of measures and rates used by insurance companies, and intended to optimize the methods by employing more scientific approaches. in order to do so, first, the factors affecting the risk of policy holders were identified and after comparing these factors with existing data in the database of surveyed company, the final factors were selected. then, after preprocessing these data, prediction of the damage class and the quantity of policyholders’ potential damages were accomplished using neural networks model. so that, with using these results and considering their damage ratio, insurance companies could define optimized premium rates for their policies. the results showed that the offered model was able to predict the damage class and potential damages of policy holders respectively with 91 and 87 percent accuracy.