Implementation of classifiers for choosing insurance policy using decision trees: a case study

In this paper, we use decision trees to establish the decision models for insurance purchases. Five major types of insurances are involved in this study including life, annuity, health, accident, and investment-oriented insurances. Four decision tree methods were used to build the decision models including Chi-square Automatic Interaction Detector (CHAID), Exhaustive Chi-square Automatic Interaction Detector (ECHAID), Classification and Regression Tree (CRT), and Quick-Unbiased-Efficient Statistical Tree (QUEST). Six features were selected as the inputs of the decision trees including age, sex, annual income, educational level, occupation, and risk preference. Three hundred insurants from an insurance company in Taiwan were used as examples for establishing the decision models. Two experiments were conducted to evaluate the performance of the decision trees. The first one used the purchase records of primary insurances as examples. The second one used the purchase records of primary insurances and additional insurances. Each experiment contained four rounds according to different partitions of training sets and test sets. Discussion and concluding remarks are finally provided at the end of this paper.