Predictive Modeling for Driver Insurance Premium Calculation Using Advanced Driver Assistance Systems and Contextual Information
Telematics devices have transformed driver risk assessment, allowing insurers to tailor premiums based on detailed evaluations of driving habits. However, integrating Advanced Driver Assistance Systems (ADAS) and contextualized geolocation data for predictive improvements remains underexplored due t...
| Autores: | , , , , |
|---|---|
| Tipo de documento: | artigo |
| Estado: | Versión aceptada para publicación |
| Data de publicação: | 2025 |
| País: | España |
| Recursos: | Universidad de Barcelona |
| Repositório: | Dipòsit Digital de la UB |
| OAI Identifier: | oai:diposit.ub.edu:2445/219131 |
| Acesso em linha: | https://hdl.handle.net/2445/219131 |
| Access Level: | Acceso aberto |
| Palavra-chave: | Assegurances d'automòbils Conducció de vehicles de motor Risc (Assegurances) Primes (Assegurances) Automobile insurance Motor vehicle driving Risk (Insurance) Insurance premiums |
| Resumo: | Telematics devices have transformed driver risk assessment, allowing insurers to tailor premiums based on detailed evaluations of driving habits. However, integrating Advanced Driver Assistance Systems (ADAS) and contextualized geolocation data for predictive improvements remains underexplored due to the recent emergence of these technologies. This article introduces a novel risk assessment methodology that periodically updates insurance premiums by incorporating ADAS risk indicators and contextualized geolocation data. Using a naturalistic dataset from a fleet of 354 commercial drivers over a year, we modeled the relationship between past claims and driving data through claims frequency using Poisson regression and claims occurrence probability using machine learning models, including XGBoost and TabNet. The dataset is divided into weekly profiles containing aggregated driving behavior, ADAS events, and contextual attributes. Risk predictions from these models are used to compute weekly premiums for each driver. SHAP is employed to interpret the machine learning model predictions. Results indicate that XGBoost achieved the lowest Log Loss, reducing it from 0.59 to 0.51 with the inclusion of ADAS warnings and driving context. However, these improvements were not consistent across all models and did not show statistically significant differences in ROC AUC values. The proposed methodology computes weekly premiums based on risk predictions from these models, penalizing risky behaviors while incentivizing safe driving behaviors. This dynamic pricing can be incorporated into the insurance lifecycle, enabling tailored policies based on emerging technologies. The study demonstrates the value of integrating diverse data sources for bespoke risk assessment and weekly insurance pricing |
|---|