GENERALIZED ADDITIVE MODEL FOR VEHICLE INSURANCE PREMIUM CALCULATION BASED ON MILEAGE AND CONTRACT DURATION

Premiums are sums of money stipulated by insurance or reinsurance companies and agreed upon by policyholders; they are payable according to the terms of insurance or reinsurance agreements. When calculating premium rates, insurance companies typically consider the risk exposure of the insured vehicle, which is crucial in estimating the number of claims. While the duration of the insurance contract is a primary factor in assessing risk exposure, other elements, such as the distance traveled, also significantly impact accident risk. This study aims to enhance the computation of risk exposure for vehicles by incorporating both the distance traveled and the duration of the insurance contract. The objective is to evaluate the combined effect of mileage and insurance contract duration on the number of claims via a generalized additive model (GAM). The GAM is chosen for its ability to capture potential nonlinear relationships between covariates and response variables. In this research, the GAM is constructed via cubic splines, and model coefficients are estimated via the penalized iteratively reweighted least squares (PIRLS) method. Upon estimating the model coefficients, the GAM is used to predict claim frequency, which can subsequently inform the relativity of premium rates compared with a reference premium. This methodology is then applied to vehicle insurance claim data to establish more accurate premium rates, considering both distance traveled and contract duration.