Periodic Reporting for period 2 - FAIR (Future Advanced system for an on-demand Insurance Reliable product based on driving behaviour analysis)
Berichtszeitraum: 2019-04-01 bis 2020-03-31
During the 2 years of the project AIR has the aim is to positively impact the insurance sector by launching in the market a platform that enables insurance product fully integrated with cars, based on telemetry and driving behaviour analysis for the implementation of a new business model.
The goal of the FAIR project is to develop and generate a predictive algorithm that improve and increase
• real-time monitoring
• improve maintenance
• increase safety
• create customized insurance policies
• reduce cost associated to insurance premiums
On the government side, the FAIR project will support the governmental authorities to improve a fundamental aspect of our society: the public security.
The platform will increase Car dealers and fleet companies’ overall marginality by:
− enhanced customer retention through tailor made aftersales services
− through predictive maintenance by reduction of the claim costs
The pay-as-you-drive insurance products, bought by Drivers/Private car owners through the App (On-demand insurance product, customized on consumers’ needs) also through recurring base payments, providing relevant savings for customers, enhancing safety and time savings.
On the insurance companies’ side, the project aim is to realize a crash algorithm that will allow them to:
• prevent fraud on claims
• predictive maintenance analysis
In a nutshell, FAIR unique selling points are:
− Discounted Usage Based Insurance
− Low risky costumer acquisition with Driving behaviour selection
− Pay-Per-Mile insurance policy
− Healthier car thanks to predictive maintenance
The goal of the FAIR project is to develop and generate a predictive algorithm that improve and increase
• real-time monitoring
• improve maintenance
• increase safety
• create customized insurance policies
• reduce cost associated to insurance premiums
On the government side, the FAIR project will support the governmental authorities to improve a fundamental aspect of our society: the public security.
The platform will increase Car dealers and fleet companies’ overall marginality by:
− enhanced customer retention through tailor made aftersales services
− through predictive maintenance by reduction of the claim costs
The pay-as-you-drive insurance products, bought by Drivers/Private car owners through the App (On-demand insurance product, customized on consumers’ needs) also through recurring base payments, providing relevant savings for customers, enhancing safety and time savings.
On the insurance companies’ side, the project aim is to realize a crash algorithm that will allow them to:
• prevent fraud on claims
• predictive maintenance analysis
In a nutshell, FAIR unique selling points are:
− Discounted Usage Based Insurance
− Low risky costumer acquisition with Driving behaviour selection
− Pay-Per-Mile insurance policy
− Healthier car thanks to predictive maintenance
During these 2 years of project the FAIR consortium has worked to fulfil the scope of the project, finalizing the activities on going and leading an overall achievement of the objectives foreseen.
The main results achieved are presented below:
Crash management flow
The crash management flow developed in the AIR platform allows the analysis of the crash events to detect real crashed occurred on test cars. A trace of the event, the Crash Report, would be generated to describing the impact recorded, the driving behaviour kept 10 mins before the crash with map positioning and the occurrence of diagnostic events after the crash.
Dealers, through the Mydesk, will be allowed to manage issues and receive notification of the occurrence of a crash on one of the car sold through them. By managing the notification, dealer can retrieve the vehicle and plan the recovery interventions needed. MyDesk can be considered as an advanced CRM platform that enhance the marginality and efficiency of of the dealer.
The Software Data Management Platform
The Data management platform, which oversees ingesting, storing, processing, and sharing data collected from IoT external sources to the other FAIR platform modules and to third party entities.
The platform is structured in several layers, the following main areas can be identified:
− Ingestion layer,
− Real time processing layer,
− Batch processing layer,
− Application layer,
− Business intelligence layer,
FAIR Platform and Government Facilities
Air started a collaboration with Lombardia and Piemonte regions on a project that aims to limit the circulation of the most polluting cars through a mileage threshold. Air is proposing to the Lombardia Region reports on the Co2 distribution based on connected cars.
Tariff algorithm
Tariff algorithm has been released a first iteration of dynamic MTPL tariff pricing based on location data (via ZIPCODE) and the inclusion in the tariff algorithm of a subset of the KBIs developed within the Data Management platform.
Behavioral machine learning
ML techniques have been applied to the Dongle data to get a segmentation of the drivers according to the driving style detected, analysing the single trips recorded and detect extreme driving styles considering as discriminating variables the extreme driving events, the duration of the events and the speed recorded.
The algorithm other output is the definition of a classifier that can classify easily the new trips coming from the devices without reperforming the whole process.
In order to have a more reliable Behavioral algorithm it is mandatory a study related to the capacity of the data to describe the risk exposure of the driver and define a link between the KBI and the risk of crash.
Digital process and Master Data Management
The FAIR project business model concern 2 main components:
• Insurance services
• IOT Service for partners and final users (driver) based on connected car data
A new business model based on a subscription economy pricing and billing concept will be implemented to sell IoT services and setting up automated billing schedules on a recurring base.
Air’s offering related to AUTOMOTIVE IoT SERVICES is based on 5 components:
1) Type of client
2) Type of product
3) Additional services:
4) On demand services
5) Payment plans
Billing and Accounting
FAIR aim to a subscription model by automating payment transactions and managing subscription-based services for its customers allowing:
• automate recurring operations
• recalculate recurring revenue streams
• reliable financial cashflow forecasting
The main results achieved are presented below:
Crash management flow
The crash management flow developed in the AIR platform allows the analysis of the crash events to detect real crashed occurred on test cars. A trace of the event, the Crash Report, would be generated to describing the impact recorded, the driving behaviour kept 10 mins before the crash with map positioning and the occurrence of diagnostic events after the crash.
Dealers, through the Mydesk, will be allowed to manage issues and receive notification of the occurrence of a crash on one of the car sold through them. By managing the notification, dealer can retrieve the vehicle and plan the recovery interventions needed. MyDesk can be considered as an advanced CRM platform that enhance the marginality and efficiency of of the dealer.
The Software Data Management Platform
The Data management platform, which oversees ingesting, storing, processing, and sharing data collected from IoT external sources to the other FAIR platform modules and to third party entities.
The platform is structured in several layers, the following main areas can be identified:
− Ingestion layer,
− Real time processing layer,
− Batch processing layer,
− Application layer,
− Business intelligence layer,
FAIR Platform and Government Facilities
Air started a collaboration with Lombardia and Piemonte regions on a project that aims to limit the circulation of the most polluting cars through a mileage threshold. Air is proposing to the Lombardia Region reports on the Co2 distribution based on connected cars.
Tariff algorithm
Tariff algorithm has been released a first iteration of dynamic MTPL tariff pricing based on location data (via ZIPCODE) and the inclusion in the tariff algorithm of a subset of the KBIs developed within the Data Management platform.
Behavioral machine learning
ML techniques have been applied to the Dongle data to get a segmentation of the drivers according to the driving style detected, analysing the single trips recorded and detect extreme driving styles considering as discriminating variables the extreme driving events, the duration of the events and the speed recorded.
The algorithm other output is the definition of a classifier that can classify easily the new trips coming from the devices without reperforming the whole process.
In order to have a more reliable Behavioral algorithm it is mandatory a study related to the capacity of the data to describe the risk exposure of the driver and define a link between the KBI and the risk of crash.
Digital process and Master Data Management
The FAIR project business model concern 2 main components:
• Insurance services
• IOT Service for partners and final users (driver) based on connected car data
A new business model based on a subscription economy pricing and billing concept will be implemented to sell IoT services and setting up automated billing schedules on a recurring base.
Air’s offering related to AUTOMOTIVE IoT SERVICES is based on 5 components:
1) Type of client
2) Type of product
3) Additional services:
4) On demand services
5) Payment plans
Billing and Accounting
FAIR aim to a subscription model by automating payment transactions and managing subscription-based services for its customers allowing:
• automate recurring operations
• recalculate recurring revenue streams
• reliable financial cashflow forecasting
"The FAIR project is configured as highly innovative due to the output provided from the platform and the target that are based on Machine learning techniques and big data analysis, and it will allow to create Usage Based Insurance (UBI) which rely upon big data analyses of drivers' needs and information.
Machine learning methods will be used to analyse the collected data and a synergic integration with social networks’ feedbacks will be implemented in order to optimize the behavioural variables.
Machine learning techniques will implement predictive results analysing 5 different levels of complexity:
1. Vertical behavioural data (Km covered, driving time, consumptions): data from connected cars
2. Contextual Data: data from third parties
3. Cost of claims: claims management
4. Telemetry data: car’s health and predicting maintenance
5. Engagement data: Notification for the driver
Big data analysis is based on different kind of algorithms:
− Driving Behavioural algorithm to understand better the real behaviour of drivers and consumers’ needs, developing appropriate mitigation solutions to enhance ""driver"" behaviour under such scenarios.
− Insurance tariff algorithm to correlate driving behaviour and insurance risk characteristics, and consequently making a new insurance policy tariff, related to driving behaviour and not only on insurance risk characteristics.
− Crash algorithm to help insurance companies in preventing fraud and weak analysis of claims. Moreover, it will predict the average cost of each claim.
Big data, and its elaboration, bring new opportunities enabling FAIR a competitive advantage in terms of:
− Risk Selection (Low risky customer acquisition)
− Risk-based Pricing (UBI )
− Services ( value added and predictive maintenance)
− Loyalty and “Behaviour engagement” ( rewarding and gamifications)"
Machine learning methods will be used to analyse the collected data and a synergic integration with social networks’ feedbacks will be implemented in order to optimize the behavioural variables.
Machine learning techniques will implement predictive results analysing 5 different levels of complexity:
1. Vertical behavioural data (Km covered, driving time, consumptions): data from connected cars
2. Contextual Data: data from third parties
3. Cost of claims: claims management
4. Telemetry data: car’s health and predicting maintenance
5. Engagement data: Notification for the driver
Big data analysis is based on different kind of algorithms:
− Driving Behavioural algorithm to understand better the real behaviour of drivers and consumers’ needs, developing appropriate mitigation solutions to enhance ""driver"" behaviour under such scenarios.
− Insurance tariff algorithm to correlate driving behaviour and insurance risk characteristics, and consequently making a new insurance policy tariff, related to driving behaviour and not only on insurance risk characteristics.
− Crash algorithm to help insurance companies in preventing fraud and weak analysis of claims. Moreover, it will predict the average cost of each claim.
Big data, and its elaboration, bring new opportunities enabling FAIR a competitive advantage in terms of:
− Risk Selection (Low risky customer acquisition)
− Risk-based Pricing (UBI )
− Services ( value added and predictive maintenance)
− Loyalty and “Behaviour engagement” ( rewarding and gamifications)"