GlowByte has developed an analytical platform for Ingosstrakh to counter fraud

2024: Anti-Fraud Analytics Platform Development

GlowByte has developed an analytical platform for Ingosstrakh to counter fraud in the field of auto insurance. Experts have implemented a set of solutions that allows you to more accurately identify fraudulent losses using machine learning models and graph analytics. This was announced on February 27, 2024 by representatives of GlowByte.

The system allows you to reduce the time to identify existing fraudulent schemes, increase the speed of identifying new methods of fraud, as well as identify potential fraudsters and prevent them from entering the company.

The solution is implemented on open source software: Camunda, Python, ArangoDB, PostgreSQL.

Automated fraud risk assessment is based on machine learning methods. By analyzing loss information, a mathematical model allows you to identify hidden patterns and statistical relationships in data, a certain combination of which indicates a high or low likelihood of fraud. During the training phase, a GINI score of 0.72 to 0.75 for various products was achieved.

The graph analysis system detects cycles of connectivity between participants, ROAD ACCIDENT connections with known fraudsters, and also calculates various business indicators of the environment in which the loss is included. For example, the presence in the environment of a client of people with a refusal to pay insurance compensation in connection with the revealed facts of fraud or a large number of losses associated with a single phone number.

This is an ambitious and large-scale project in the Russian insurance market, which we managed to implement exclusively on open source software. source code This system allows the customer to reduce the time for identifying complex fraudulent schemes from several days to several hours, as well as timely identify hidden fraudulent schemes. This makes the business more efficient and stable, "said Evgeny Chernoburov, head of insurance practice at GlowByte.

{{quote 'The data analysis system now allows you to see implicit links and patterns that until then went unnoticed. This makes it easier to identify suspicious actions and prevent financial losses for Ingosstrakh. The system opens up new prospects and opportunities for the effective use of graph analytics in various areas of professional activity, - said Ivan Kotlyarovsky, project manager of the retail business settlement department at Ingosstrakh. }}