I wonder about approaches for a risk segmentation applied to auto insurance. i.e.
Suppose I have all the states within United States, and I wanna say: "Ok, Alaska and Alabama belong to group 1; Connecticut, Delaware and California belong to group 2, some zip codes from Arkansas and Colorado belong to group 3, and so on" but
- I wanna create this segmentation based on several variables (sinistrality, crime rate, economic factors, etc.) and
- I want a "smooth segmentation" i.e., If i live at "Street 1" belonging to group 1 and I cross to "Street 2" from group 2 I wanna avoid getting a premium risk twice higher than the premium I've got at "Street 1". I mean, if this segmentation were plotted i'd get a sort of "gradient"
Any suggestion will be appreciated