On Application Of Data Mining In Incentive Mechanism Of Life Insurance Agent

Life insurance is an important branch of the insurance industry. Data miningtechniques have been widely applied to problems in life insurance. The application ofdata mining for life insurance has important practical significance. Fiercercompetitions will face our life insurance enterprise since home market is to be broadlyopened to other countries. Life insurance companies generally lack informationfeedback and performance analysis in agent’s incentive system and agent’s behavior.Therefore, it becomes increasingly important to improve the incentive mechanism oflife insurance agent by using data mining techniques. At present, there is blindnessand bad real-time performance in the selection of incentive modes, which could notmeet the needs on the development of the life insurance company. Meanwhile, one ofthe other reasons must be the insufficient analysis to the profit analysis of theincentive decision. This paper focuses on the application of the agent incentivemechanism research in China’s life insurance industry. Based on some main datemining technologies, this paper focuses on the decision tree methods and clusteringalgorithms according to the data of agents’ incentive form information from lifeinsurance companies. The main works of the paper are as follows:Many factors influence the decision-making of incentive forms. Life insurancecompanies need to avoid one-sided decisions and tedious process. Therefore, thispaper proposed a decision tree model for life insurance agent based incentives. C4.5Tree and Random Tree are used to identify the factors that influence the result ofincentives forms. And then, this paper optimizes the model by setting up theperformance to get better goodness of model fitting. Eventually, comparing theprediction results with actuarial values, we find that the established decision treemodel has shown a satisfying performance. F-Measure of proposed method is86.6%.Furthermore, aiming at solving the benefit evaluation of incentives, this paperconstructs a cluster model and selects information indicators of incentive benefits.Clustering analysis is done by WEKA software. Hierarchical clustering method ischosen to construct significant life insurance company’s performance categoriesbecause of its better performance than K-means clustering. The conclusions canprovide effective suggestions to optimum adjust the proportion of the incentive forms.