Cinimex has implemented an ML project for Rosgosstrakh

2022: Development of Risk and Loss Services

On April 7, 2022, the Cinim ex company announced that, together with Rosgosstrakh, it had implemented a comprehensive project using the latest data assessment methods. Data analysts have developed machine learning services that allow the insurer to more accurately assess risks, predict large losses, and assess portfolios of agents, brokers and partners.

The Cinimex team developed a geographic segmentation model. The algorithm optimizes risks by working with selected customer segments and geozones. Historical loss data and territorial characteristics from open sources are used for risk assessment. Clustering takes into account free zones, where the insurance company is poorly represented and has the potential for sales growth. In the future, it is possible to expand the analysis due to demographic and other statistical data.

The use of artificial intelligence allows you to qualitatively assess the portfolios of agents, brokers and other partners cooperating with the company. For example, a change in agent behavior and the execution of non-standard contracts may indicate an increase in risk. To predict unprofitability, the system estimates and predicts sales potential for various periods.

The peculiarity of the project is in the segmentation process itself, both according to numerical source data (many sources) and binding to the geolocation of each client (geosegmentation). Different machine learning models are responsible for each of the subtasks, which should work effectively not only autonomously, but also in conjunction, "said Rodion Martynov, Project Manager of Cinimex.

The choice of technology was influenced by high competition and an ever-expanding range of products in the financial sector. An important factor was the increase in the number of customers, and the increase in data for each of them. We turned to Cinimex for development from scratch, because we had goals that existing boxed solutions simply do not achieve. Most boxed solutions do not take into account all data and cannot be fully adapted to the specifics of our company (for example, predictive analysis based on map data and graphs). Within the framework of this project, we needed a project development that was originally based on company data and specific business rules. We were also convinced that the decision to use in the project for the most part open source software and independent development was correct, taking into account the trend for import substitution, - said Olga Veresova, Head of the Analysis and Control Department of the Rosgosstrakh insurance company.