Challenge

Monitoring driver behaviour offers significant financial benefits for insurance companies. Traditionally, this has required installing tracking devices in vehicles to transmit data on speed, location, and other driving metrics—a process that is both costly and cumbersome for customer onboarding. The insurance company sought a more efficient solution by leveraging mobile devices to capture driver behaviour and more accurately price their insurance products.

Our Approach

We implemented a statistical filtering method to estimate key driving metrics using data observed from mobile phones. This time series approach analysed the phone’s motion in relation to the car’s speed and location. Outputs from the model included critical factors such as vehicle orientation, phone usage, texting, and other behaviours, which were then incorporated into the premium pricing model.

Outcomes

• 3D Visualisation: Developed dashboards that provided the client with 3D reconstructions of vehicle orientation and mobile phone positioning during any trip.

• Cloud-Based Deployment: Deployed a production-ready system on AWS infrastructure, ensuring seamless integration with the client’s existing systems.

• Real-Time Monitoring: Enabled near real-time updates for significant driving events, such as harsh braking and sudden acceleration, enhancing the accuracy of risk assessments and premium pricing.