Centro de Documentacion de Fundación MAPFRE - Using Kohonen's self-organizing feature map to uncover automobile bodily injury claims fraud

MAP20071023781

Using Kohonen's self-organizing feature map to uncover automobile bodily injury claims fraud / Patrick L. Brockett, Xiaohua Xia, Richard A. Derrig

31 h. ; 30 cm

Ponencia presentada en American Risk and Insurance Association Annual Meeting, Seattle, 1995

Sumario: Claims fraud is an increasingly vexing problem confronting the insurance industry. In this empirical study, the authors apply Kohonen's Self-Organizing Feature Map to the classification of automobile bodily injury (BI) claims by the degree of fraud suspicion, each claim being classified into one of four classes with different levels of suspected fraud. Feedforward neural networks and a backpropagation algorithm are used to investigate the validity of the Feature Map approach. A comparative experiment illustrates the potential usefulness of the proposed methodology. They show that this technique performs better than both an insurance adjuster's fraud assessment and an investigator's fraud assessment with respect to consistency and reliability, but that the extent of misclassification of the model remains high on the data sample

1. Seguro de automóviles . 2. Reclamaciones . 3. Fraude en el seguro . 4. Valoración de daños . 5. Métodos analíticos . 6. Métodos experimentales . 7. Conferencias . 8. Congresos . 9. Fraude . I. Xia, Xiaohua . II. Derrig, Richard A. . III. American Risk and Insurance Association (1995: Seattle). Annual Meeting . IV. Título.