This article introduces a simulation engine for adaptive telematics (SEAT), which flexibly generates insurance claims datasets from driver telematics information that matches the specific profile of a target market. Generating adaptive telematics data via SEAT is a two-stage process. In the first stage, SEAT uses predetermined distributions of traditional policy characteristics from a target market as inputs and replicates these policy characteristics based on their distributions. In the second stage, SEAT generates the remaining covariates and insurance claims accordingly, based on configurations of the traditional policy characteristics with possible perturbations. We illustrate how SEAT generates an adaptive telematics dataset to match the South Korean insurance market and compare its behavior with the source telematics dataset on which the algorithm is based. We hope that both practitioners and researchers will use this publicly available simulation engine (https://github.com/bheeso/SEAT.git) and adaptive datasets to explore the usefulness of driver telematics data for developing diverse models of usage-based insurance.
So et al. (Thu,) studied this question.