Customer profitability forecasting using Big Data analytics: A case study of the insurance industry

Abstract

This paper proposed a new customer profitability method for the insurance industry by adding liability reserve. Considering the historical purchasing behavior and the foreseeable future cash flow, the proposed method can measure the real insurance customer contribution effectively. In addition, this paper firstly applies random forecast regression, a method for Big Data analytics, to forecast insurance customer profitability. Comparing with other models, we find that random forest outperforms traditional forecasting methods, such as linear regression, decision tree, SVM and generalized boosted model. Empirical study finds that customer’s region, age, insurance status, sex and customer source are the most important factors to predict insurance customer profitability.