I'm about to start a commercial fleet glim and I wanted to pick the brains of some of the experts on this forum. I've worked on retail motor GLIMs before with lots of data and this is my first commercial fleet glim. My queries are outlined below: Data Volume The size of the fleet account I'm working on is much smaller than the retail motor portfolio I worked on before. Is there a standard rule on the minimum amount of data (in terms of vehicle exposure years and claim numbers) a level of factor should have for the results to be considered statistically reliable? Is it enough to have significant betas or can significant betas exist with "insufficient" data volume? Can anyone recommend any good reading sources for this? The period of data I'm using is 6 years so when I carry out a time- consistency check the data is split by year. I'm just concerned as the actuary before me included loads of factors in the model which I would have excluded due to sparse data. Rating Factors What are the best/most powerful factors you would consider in a fleet model? Industry, vehicle type, area, size of fleet? The last company I worked for didn't collect data on driver details in a fleet, just number of drivers. This information is collected for this data set- is it worthwhile to model driver factors in a fleet glim as for retail motor? Types of Models Is it worth modelling by peril for fleet as for retail motor? If anyone can recommend any good sources on this it would be great!
