Classifying drivers using electronic logging devices

Abstract

In the era of personalization, being able to determine the risk of individual drivers and hence provide suitable insurance coverage to them would be a logical step. This paper proposes risk scoring for motor insurance using logged data of the drivers that are collected electronically. The proposed method uses machine learning to create a model that can be applied using the logged data. Initial studies conducted were able to achieve up to an accuracy of 79.4%. With further improvement, it can provide a suitable individual risk scoring for insurance premium computation.