A specific feature of insurance data is the existence of correlations between variables: on average, young drivers have old cars, ladies have smaller cars…
So finding which profiles are over or underpriced requires more sophisticated techniques than a simple analysis of loss ratios. Tricast
uses linear regressions, which take into account these correlations. This technique allows a more relevant analysis of the over and under pricing.
Creating rating structures or revising existing ones is just one step of a larger process. After that, it is necessary to simulate the impact of such changes. That can also be done using Tricast