Insurance Claims Risk Modeling and Forecasting Using Mathematical Models and Scorecards

Source abstract

In this paper, several car insurance claims problems are analyzed and solved via existing statistical models implementation for real-world datasets. The first problem which was studied is the problem of measuring the probability of a claim for a specific policy. This problem is solved by using a set of families of generalized linear models with an additional approach to analyze data by utilizing survival models. The best generalized linear model is then chosen according to statistical criteria. The second problem considers distinct classes of policies. A number of claims and prices are forecasted for the different groups. Same approach as for the first problem, generalized linear models are used and the best model is chosen according to statistical criterion. The third problem is the problem of scorecard generation. A brief interpretation and result of the built scorecard is also provided.