Study on the transform method of estimating discrete frequency from continuous variable: ratemaking for car repair insurance based on SAS system coding

Some discrete variable such as frequency cannot be estimated when a limited sample is drawn from the population which is not sufficient enough to represent the whole population. But the procedure records data from finite samples that can be converted to frequency estimates through computer intensive calculations. The aim of this paper is to develop an Accelerated Failure Time model for the continuous variables such as survival times or the first breakdown mileage by embedding Weibull distribution into a GLMs structure. Then we can derive the hazard ratio function and transform the continuous variable modeling into discrete variable inference. A numerical illustration based on a data derived from a Chinese auto dealer is performed with the statistical software SAS. The rate was made for different types of vehicles and their different parts based on the minimum repairing and the purchase of car repairing insurance for a certain time.