Big Data for Insurance: How Algorithms Affect Car Insurance Prices

According to the IDC, international analysis and advisory company, in 2016, the global business analytics software and Big Data market was valued at 130 billion dollars, which is 11 per cent higher than last year. As analysts predict, sales of Big Data solutions will be growing up during the next three years. These days, such technologies are mostly in demand for banks, telecommunication and public service companies, as well as insurance establishments.

Yaroslav Polin, student of the Software Development Department, presented his project called "The applications of Big Data in the field of insurance" at the annual contest organized by "Ingosstrakh" Insurance company. Yaroslav analyzed the applications of Big Data solutions to anti-fraud measures and personalized options for vehicle insurance.

Yaroslav Polin

"I've analyzed existing swindler cases, the information about which is saved in security archives of companies. In big companies, specialists in this field analyze both scattered data so as to look for common features, and the whole amount of transactions in order to find potential frauds. That’s how Visa works — they monitor all transactions online so as to find suspicious ones. Thanks to such algorithms, the company saves over 2 billion dollars annually," says Yaroslav Polin.

Usually, when dealing with data analysts, one has no access to personal information. However, algorithms help recognize customers even when names, second names and gender are unknown.

"The data arrays used by bank analysts include information about expenses and places where items were bought. Using this information, specialists create algorithms and then apply them. As for gender detection, sometimes a system can give over 90% of correct answers, but such tasks as purchasing behavior forecast for the next month are much more difficult," says the student.

When taking part in hackathons and solving real tasks, analysts deal with various cases and problems. Sometimes, specialists get unexpected and irrational results. For instance, having analyzed a large amount of data related to vehicle insurance, one can suddenly learn that owners of red cars get into car accidents more frequently than others. This seems to be illogical; however, in fact, such correlations can affect price formation at insurance companies, thereby helping avoid losses.

For many years, Russian IT holdings like Mail.Ru Group or Rambler have been conducting Big Data analyses by themselves. There are also such data mining facilities as Yandex Data Factory, which offers data processing services for large businesses, as well as small companies and startups. According to Yaroslav Polin, these technologies are only in the formative period. Yaroslav continues to work on this field and develop his project.

"Next year, I’d like to take part in projects of Ingosstrakh including both analytical work and dealing with real data arrays. If a company is ready to give some real information, I will participate with a great pleasure. I think that this collaboration will be useful for both me and the company," summed up YaroslavPolin.