A fuzzy telematics data-driven approach for vehicle insurance policyholder risk assesment

摘要

Recent technological advances in telematics devices are providing companies with opportunities to offer new products and services to their customers. Many insurance companies are exploiting the value of telematics devices to develop new insurance models based on use. To help these companies assess the risk associated with policyholders, we have developed a new fuzzy approach that analyzes complex driving patterns using big telematics data. A fuzzy risk score is calculated for each driver according to the frequency of their risky driving behaviors and the average speed of individual trips. The aggregate of these scores reflects the risk level of each policyholder and, thus, the potential exposure for insurers. The approach has multiple benefits for insurance companies, particularly in setting appropriate premiums for individual policyholders to counter some of that risk. We tested the proposed approach on a dataset containing the driving habits of more than 2500 drivers. The proposed approach can estimate the risk level of each policyholder based on their driving features.