Investigating driver behaviour effects on usage based car insurance focusing on accident prediction

Abstract

This study looks into the impact of driver behaviour on accident prediction in the context of Usage-Based Insurance (UBI), a system that uses telematics data to adjust insurance premiums based on real-time driving habits. The study hypothesizes that UBI systems can successfully incentivize better driving by directly linking insurance costs to driving behaviour, resulting in fewer accidents. Using the Cox proportional hazards model, the study investigates a mix of telematics and accident data to identify significant behavioural traits that affect accident risk. The findings indicate that features such as acceleration patterns, braking behaviour, and speeding are major predictors of accidents, shedding light on more accurate premium pricing strategies under UBI. The investigation, which was mostly based on overseas data due to restricted access to local data, highlights UBI's potential to improve road safety and insurance pricing justice. However, the findings indicate that localized data is critical for fine-tuning UBI initiatives in emerging economies such as Kenya. The paper concludes with recommendations for future research that incorporates regional data, which will aid in continuing and refining these insights for wider application.

Key words: Usage Based Insurance, Driver behaviour, Telematics data, accident prediction, cox proportional hazard model