MODELLING CLAIM FREQUENCY AND LOSS DUE TO CLAIMS OF AUTOMOBILE INSURANCE

The purpose of this study is to model the claim frequency and loss due to claims of automobile insurance. The Poisson, Gamma and their offset termed regression models were applied on the motor insurance data as compiled by the Swedish committee on the analysis of risk premium. Using the Generalised Linear Models (GLM), the applied models were diagnosed and the results of the application was obtained, analysed and the better model identified. The fitted models were further used to carry out predictions of claim frequency and severity (loss due to claims), where the offset termed Poisson model was seen to predict closer than the normal Poisson model. For the prediction of loss due to claims (payments), the Gamma model without the offset term is seen to perform better.