Challenge
A peer-to-peer motorcycle rental platform was looking to use DigiSure's Risk Control product to predict insurance claims risk and reduce overall insurance losses on their platform. Furthermore, they were interested in understanding what specific variables increase the probability of a claim occurring.
Solution
To add precision to Risk Control, DigiSure's data science team created a machine learning model predicting claims risk by combining customer data, third party data sources and claims data to identify features that were predictive of insurance losses. One of these features is the ratio of an individual's height to the seat height of a motorcycle. Upon looking at DigiSure's claims data, the head of claims found that a high percentage of claims occur when there is a 'tip-over'—that is, a slow moving vehicle that loses stability and falls over. This 'tip-over' scenario occurs much more frequently when an individual is shorter relative to the seat height of the motorcycle. When the data science team tested this rule in the model, they were able to find that a driver height to seat height ratio below 2 increased the probability of a claim by 20%.
Impact of the Model
By combining the seat height rule with over 100 different variables, DigiSure found that the riskiest 5% of drivers in our system comprise over 40% of claims. In utilizing end-to-end data sources, DigiSure is able to reduce losses and improve safety on Twisted Road's platform by identifying risky drivers.
Eager to learn how DigiSure can help your business? Request a demo to learn more.