Synthetic Dataset Generation of Driver Telematics

This article describes the techniques employed in production of a synthetic dataset driver telematics emulated from similar real insurance dataset. The generated has 100,000 policies that included observations regarding driver’s claims experience, together with associated classical risk variables and telematics-related variables. work is aimed to produce resource can be used advance models assess risks for usage-based insurance. It follows three-stage process while using machine learning algorithms. In first stage, portfolio space feature applying an extended SMOTE algorithm. second stage simulating values number as multiple binary classifications feedforward neural networks. third aggregated amount regression networks, set resulting evaluated by comparing datasets when Poisson gamma are fitted respective data. Other visualization data summarization remarkable statistics between two datasets. We hope researchers interested obtaining calibrate or algorithms will find our ot valuable.