Data mining for selection of insurance sales agents

The insurance industry of Hong Kong has been experiencing steady growth in the last decade. One of the current problems in the industry is that, in general, insurance agent turnover is high. The selection of new agents is treated as a regular recruitment exercise. This study focuses on the characteristics of data warehousing and the appropriate data mining techniques that can be used to support agent selection in the insurance industry. We examine the application of three popular data mining methods – discriminant analysis, decision trees and artificial neural networks – incorporated with a data warehouse to the prediction of the length of service, sales premiums and persistence indices of insurance agents. An intelligent decision support system, namely Intelligent Agent Selection Assistant for Insurance, is presented, which will help insurance managers to select quality agents by using data mining in a data warehouse environment.