Predicting car insurance policies using random forest

Data mining has been recently used in the field of car insurance to help the insurance companies in predicting the customers´ choices in order to provide more competitive services. In this composition, the random forest was used to develop a classification model that could be applied in predicting which of the insurance policies would likely to be chosen by the customers. The performance of the developed model was compared to several data mining techniques such as ZeroR classifier, Simple Logistics Function, Decision Tree and Naïve Bayes on a dataset contains 7 different policies. The results showed that the random forest was the most precise technique with an overall accuracy of 97.9 %.