Research on Recommendation of Insurance Products Based on Random Forest

With the rapid development of recommendation system, how to predict user's behavior accurately become more and more important. In this paper, random forest is applied to recommend insurance products and compared with ID3, C4.5, Nave-Bayes and Nearest-neighbor. Experiment results show that the prediction error of random forest is 2.02% lower than ID3, 1.09% lower than C4.5, 1.67% lower than Nave-Bayes and 5.97% lower than Nearest-neighbor. Therefore, it is highly feasible to recommend insurance products with random forests.