Artificial neural network (ANN) with back propagation algorithm (BP) and ANFIS was chosen as an alternative technique in modeling motor insurance claims. In particular, an ANN and ANFIS technique is applied to model and forecast the Malaysian motor insurance data which is categorized into four claim types; third party property damage (TPPD), third party bodily injury (TPBI), own damage (OD) and theft. This study is to determine whether an ANN and ANFIS model is capable of accurately predicting motor insurance claim. There were changes made to the network structure as the number of input nodes, number of hidden nodes and pre-processing techniques are also examined and a cross-validation technique is used to improve the generalization ability of ANN and ANFIS models. Based on the empirical studies, the prediction performance of the ANN and ANFIS model is improved by using different number of input nodes and hidden nodes; and also various sizes of data. The experimental results reveal that the ANFIS model has outperformed the ANN model. Both models are capable of producing a reliable prediction for the Malaysian motor insurance claims and hence, the proposed method can be applied as an alternative to predict claim frequency and claim severity.
[1]
Vladik Kreinovich,et al.
Soft computing in measurement and information acquisition
,
2003
.
[2]
Siti Mariyam Hj. Shamsuddin,et al.
Data Modeling for Kuala Lumpur Composite Index with ANFIS
,
2008,
2008 Second Asia International Conference on Modelling & Simulation (AMS).
[3]
Tshilidzi Marwala,et al.
Improving the Performance of the Ripper in Insurance Risk Classification - A Comparitive Study using Feature Selection
,
2011,
ICINCO.
[4]
Mehdi Khosrowpour,et al.
Annals of Cases on Information Technology
,
2002
.
[5]
Kate Smith-Miles,et al.
Modelling the Effect of Premium Changes on Motor Insurance Customer Retention Rates Using Neural Networks
,
2001,
International Conference on Computational Science.
[6]
Jie Lu,et al.
Intelligent financial warning model using Fuzzy Neural Network and case-based reasoning
,
2011,
2011 IEEE Symposium on Computational Intelligence for Financial Engineering and Economics (CIFEr).