insurance claim classification: a new genetic programming approach

In this study we provide insurance companies with a tool to classify the risk level and predict the possibility of future claims. the support vector machine (svm) and genetic programming (gp) are two approaches used for the analysis. basically, in iran insurance industry there is no systematic strategy to evaluate the car body insurance policy. companies refer mainly to the world experience and employ it to rate the premium. an insurance claim dataset provided by an iranian insurance company with a sample size of 37904 is considered for programming and analysis. according to the structure of the dataset, a supervised learning algorithm was used to describe the underlying relationships between variables. the model accuracy is over 90% and the outcomes indicate that car type, car plate, car color and car age were the main four factors contributing in prediction of claims.