May 30, 2017 at 12:25How Big Data Can Be Used in Insurance: ITMO University Projects
According to IDC, in just 3 years, the market for specialized software that works with Big Data could reach $ 203 billion. Now it is estimated at 130 billion, but the demand from the banking sector, insurance and telecommunications companies is only growing.
Today we will talk about what analytical tasks at the junction of the insurance and "big data" spheres are solved by ITMO University projects. Flickr / Richard Masoner / CC
The main interest of companies and their customers is effective risk management and cost optimization. In the direction itself, there is nothing new - risks and the desire to preserve a particular property, financial assets or other values are central elements of the entire insurance industry.
Forecasting and assessment are an integral part of the risk management process. Compared with the times when the first insurance companies were created (for example, following the results of the Great Fire in London in 1666), the degree of complexity of the valuation methods and the volumes of the data analyzed increased many times over.
The advent of “smart devices” and new opportunities for working with huge amounts of data transforms the tasks that face the analytical departments of insurance companies. Starting with the modernization of existing infrastructure for collecting and analyzing information and to revising the basic approaches to customer service. All this is necessary not only to reduce the cost of insurance, but also to increase the transparency of data collection processes - understanding by the client of what information he provides and what he receives in return.
To discuss such problems, ITMO University uses platforms such as the iDealMachine accelerator. One of the iDoMa accelerator projects involves monthly seminars for everyone who is interested in the startup industry. As part of one of these events, we held a meetingwith representatives of leading insurance companies with whom workshop participants discussed the lack of startups in the insurance industry.
As one of the main reasons for this phenomenon, experts have identified too accentuated bias of insurance companies in favor of their subsidiaries. Today's startups in this area are mainly spin-offs of large insurance companies and do not want to position such projects as a separate business. On the other hand, the interest of insurance companies in working in the in-house format can attract those who are willing to sell their project or make it part of a corporation.
In 2008, the journal Nature published one of the first definitions of Big Data, which suggested the existence of special methods and tools for processing huge amounts of information. The current reputation of Big Data is largely spoiled by false expectations and the search for a universal solution that will allow you to move to a completely different technological level.
The problem is not so much in the popularity of the term “Big Data”, but in the not quite accurate statement of tasks and the choice of tools for solving them. The bulk of the projects that deal with Big Data comes to “big disappointments” only because it cannot formulate the exact purpose of such work. One way or another, many are still convinced that Big Data is nothing more than a marketing ploy.
Flickr / ted eytan/ CC
A similar situation exists in the insurance industry. To be more precise, in auto insurance - in the field of application of telematics devices to track driving style and assess the risks of the occurrence of a particular insurance event. Our market boasts only 50 thousand personalized offers that have been implemented by insurance companies over the past few years using telematics. Statistics show that these sales volumes make up no more than 1% of the total number of all CASCO policies.
There can be many reasons: from the reluctance to change the driving style to a less aggressive one, to the fears of the unfair use of personal data (almost all motorists have encountered this). The main problem here is that in Russia there is no specialized legislative base that could standardize such systems and the procedure for working with them for insurance companies. Nevertheless, insurers do not lose hope and work out various options for technological solutions that will allow the market to become interested in real results, and not serve solely as a marketing bait.
Another area of insurance in which the use of Big Data is slowly but surely gaining momentum is property insurance. Built-in sensors help to reduce risks here, which allow you to detect unauthorized entry into the room, gas leak, breakthrough of one or another component of the water supply system or fire in time. Such devices are of interest not only from the point of view of minimizing losses (when something went wrong), but also serve as the basis for the formation of a more favorable insurance offer in combination with statistics and the accumulated history of insurance cases.
The life and health insurance segment is developing in a similar way - it has its own approaches to monitoring the status of the insured. Some of the leading offers from insurance companies take into account not only the medical history of the client, but also data from wearable devices. Thus, the client receives substantial discounts on the exchange of data that is accumulated in the bases of the insurance company for further processing. Nevertheless, telematics is still the leader in terms of distribution among insurers - and most insurance data analysis projects somehow turn out to be connected specifically with automobiles.
One of our students took advantage of the interest of insurance companies in innovative developments and adapted their research work to the format of the annual contest of Ingosstrakh company. In his competitive work “Big Data in Insurance - Areas of Application”, Yaroslav Polin, a student of the Department of Software Systems at ITMO University, highlighted the most promising anti-fraud approaches and methods for personalizing the prices of OSAGO policies using the capabilities that Big Data provides.
Yaroslav conducted an analysis of current fraudulent situations and identified parameters that allow you to track suspicious messages from clients. This approach is largely reminiscent of the methods that financial institutions use to search for suspicious transactions online.
Telematics vendors, who find their benefits in working with Big Data, are also guided by this logic. The authors of one of these projects are Maxim Savelyev and Samuel Gorelik. In short, we are talking about a hardware-software complex that allows you to track various characteristics of vehicles. Based on these data, it is possible to assess the quality of driving, which cannot but arouse interest among transport companies (control of the fleet and the work of employees) and insurance companies (providing discounts for accurate driving).
The main purpose of telematic systems is to provide source data for creating personalized insurance offers. It takes into account such parameters as: vehicle speed, acceleration, braking, fuel consumption and other nuances that allow you to build a "profile" of the driver.
The developers of this vehicle monitoring system designed not only hardware, but also special software - the complex is currently being tested on several buses of the Passenger Passenger Transport St. Petersburg. Subsequently, such a system for assessing the quality of transportation can become part of the smart transport network of any city. Research in this area and the field of road quality assessment is also conducted by ITMO University staff.
Today we will talk about what analytical tasks at the junction of the insurance and "big data" spheres are solved by ITMO University projects. Flickr / Richard Masoner / CC
What are we talking about
The main interest of companies and their customers is effective risk management and cost optimization. In the direction itself, there is nothing new - risks and the desire to preserve a particular property, financial assets or other values are central elements of the entire insurance industry.
Forecasting and assessment are an integral part of the risk management process. Compared with the times when the first insurance companies were created (for example, following the results of the Great Fire in London in 1666), the degree of complexity of the valuation methods and the volumes of the data analyzed increased many times over.
The advent of “smart devices” and new opportunities for working with huge amounts of data transforms the tasks that face the analytical departments of insurance companies. Starting with the modernization of existing infrastructure for collecting and analyzing information and to revising the basic approaches to customer service. All this is necessary not only to reduce the cost of insurance, but also to increase the transparency of data collection processes - understanding by the client of what information he provides and what he receives in return.
To discuss such problems, ITMO University uses platforms such as the iDealMachine accelerator. One of the iDoMa accelerator projects involves monthly seminars for everyone who is interested in the startup industry. As part of one of these events, we held a meetingwith representatives of leading insurance companies with whom workshop participants discussed the lack of startups in the insurance industry.
As one of the main reasons for this phenomenon, experts have identified too accentuated bias of insurance companies in favor of their subsidiaries. Today's startups in this area are mainly spin-offs of large insurance companies and do not want to position such projects as a separate business. On the other hand, the interest of insurance companies in working in the in-house format can attract those who are willing to sell their project or make it part of a corporation.
Some more criticism
In 2008, the journal Nature published one of the first definitions of Big Data, which suggested the existence of special methods and tools for processing huge amounts of information. The current reputation of Big Data is largely spoiled by false expectations and the search for a universal solution that will allow you to move to a completely different technological level.
The problem is not so much in the popularity of the term “Big Data”, but in the not quite accurate statement of tasks and the choice of tools for solving them. The bulk of the projects that deal with Big Data comes to “big disappointments” only because it cannot formulate the exact purpose of such work. One way or another, many are still convinced that Big Data is nothing more than a marketing ploy.
Flickr / ted eytan/ CC
A similar situation exists in the insurance industry. To be more precise, in auto insurance - in the field of application of telematics devices to track driving style and assess the risks of the occurrence of a particular insurance event. Our market boasts only 50 thousand personalized offers that have been implemented by insurance companies over the past few years using telematics. Statistics show that these sales volumes make up no more than 1% of the total number of all CASCO policies.
There can be many reasons: from the reluctance to change the driving style to a less aggressive one, to the fears of the unfair use of personal data (almost all motorists have encountered this). The main problem here is that in Russia there is no specialized legislative base that could standardize such systems and the procedure for working with them for insurance companies. Nevertheless, insurers do not lose hope and work out various options for technological solutions that will allow the market to become interested in real results, and not serve solely as a marketing bait.
Another area of insurance in which the use of Big Data is slowly but surely gaining momentum is property insurance. Built-in sensors help to reduce risks here, which allow you to detect unauthorized entry into the room, gas leak, breakthrough of one or another component of the water supply system or fire in time. Such devices are of interest not only from the point of view of minimizing losses (when something went wrong), but also serve as the basis for the formation of a more favorable insurance offer in combination with statistics and the accumulated history of insurance cases.
The life and health insurance segment is developing in a similar way - it has its own approaches to monitoring the status of the insured. Some of the leading offers from insurance companies take into account not only the medical history of the client, but also data from wearable devices. Thus, the client receives substantial discounts on the exchange of data that is accumulated in the bases of the insurance company for further processing. Nevertheless, telematics is still the leader in terms of distribution among insurers - and most insurance data analysis projects somehow turn out to be connected specifically with automobiles.
What ITMO University does
One of our students took advantage of the interest of insurance companies in innovative developments and adapted their research work to the format of the annual contest of Ingosstrakh company. In his competitive work “Big Data in Insurance - Areas of Application”, Yaroslav Polin, a student of the Department of Software Systems at ITMO University, highlighted the most promising anti-fraud approaches and methods for personalizing the prices of OSAGO policies using the capabilities that Big Data provides.
Yaroslav conducted an analysis of current fraudulent situations and identified parameters that allow you to track suspicious messages from clients. This approach is largely reminiscent of the methods that financial institutions use to search for suspicious transactions online.
Telematics vendors, who find their benefits in working with Big Data, are also guided by this logic. The authors of one of these projects are Maxim Savelyev and Samuel Gorelik. In short, we are talking about a hardware-software complex that allows you to track various characteristics of vehicles. Based on these data, it is possible to assess the quality of driving, which cannot but arouse interest among transport companies (control of the fleet and the work of employees) and insurance companies (providing discounts for accurate driving).
The main purpose of telematic systems is to provide source data for creating personalized insurance offers. It takes into account such parameters as: vehicle speed, acceleration, braking, fuel consumption and other nuances that allow you to build a "profile" of the driver.
The developers of this vehicle monitoring system designed not only hardware, but also special software - the complex is currently being tested on several buses of the Passenger Passenger Transport St. Petersburg. Subsequently, such a system for assessing the quality of transportation can become part of the smart transport network of any city. Research in this area and the field of road quality assessment is also conducted by ITMO University staff.