Risk assessment and data-driven insurance pricing models for connected and automated vehicles

The proliferation of connected and automated vehicles (CAV) introduces several challenges in the vehicular ecosystem. One of its significant impacts is on the automobile insurance industry, as the crash risk is expected to change with safer emerging technologies. Traditionally, vehicle insurers have been proposing data-driven models to provide bespoke premiums based on how much and safely insureds drive. Nevertheless, there exists a lack of research that incorporates the safety benefits of emerging vehicular technologies in the insurance model. This thesis contributes novel methodologies to fill this gap, linking CAV data, road safety, and risk analysis. The thesis begins with a systematic review of vehicular datasets, CAV safety effectiveness, and data-driven motor insurance products. Then, based on empirical analyses of naturalistic driving datasets, it studies the predictive power of advanced driver assistance systems (ADAS) and the contextual environment where people drive (i.e., the driving context) from risk and driver behaviour perspectives. Different machine learning and statistical techniques are investigated, and explainable artificial intelligence is proposed for risk analysis models. Ultimately, the thesis contributes a methodology for motor insurers willing to incorporate information from ADAS, the driving context, and vehicle dynamics into their pricing models. The results quantify ADAS's safety benefits and highlight the predictive power of contextual attributes and ADAS warnings for risk assessment. Furthermore, studying the crash risks of different driving contexts allows for promoting countermeasures to mitigate the most severe road crashes. The contributions of the thesis are relevant for motor insurers and stakeholders in road safety and transportation domains studying the transition to CAV. Such contributions aim to incentivise, through fair insurance, the incorporation of vehicular technologies capable of promoting safer driving habits and reducing road crashes.