Automated Advisory in Car Insurance
Car insurance is a well-developed and competitive product in the insurance market. In this type of insurance, ranking of both policyholders and cars based on their assessed risk is crucial. These rankings are costly to be obtained and the underwrites mostly rely on experts' opinion for this task. In this project, we helped an insurance company automate the experts' opinion in ranking cars and policyholders.
We created a three-group ranking of policyholders using state-of-the-art unsupervised learning methods. The ranking based on this approach was partly in line with the previously known figures but has improved upon the experts' opinion. The proposed method will be eventually applied in the back-end analysis, as well as the design of the front-end platform of the company.