It was necessary to supplement the frequency and average loss models with additional factors that are based on information about the policyholder's address (distances, population of cities and their derivatives)
Stages of work: 1. Data collection 2. Generating and testing solutions, finding the right one 3. Building a Prototype Model 4. Configure Infrastructure
Additional useful factors were calculated. They are added to existing algorithms for constructing forecast frequency and average loss values. It was possible to significantly improve the models.
Areas of use
The case is useful for insurance companies. This product allows you to better balance the portfolio using cheap and open data.
We're in touch!
Email us for cooperation or if you have any questions.