Evaluating the Effectiveness of Link Functions in Gamma Regression for Estimating Vehicle Insurance Claim Costs

This study investigates the application of the Gamma regression model—a type of Generalized Linear Model (GLM)—to the analysis of insurance claim costs where the response variable is continuous, strictly positive, and right-skewed. Using real-world data from vehicle insurance claims, the research aims to assess the influence of key factors, including policyholder age, vehicle age, and vehicle type, on the average claim amount. The analysis was performed using SPSS, where both logarithmic and inverse link functions were employed to compare model fit and performance. Results revealed that the log link function provided superior model adequacy, as indicated by lower deviance and Pearson Chi-Square statistics. All independent variables showed statistically significant effects on claim costs, reinforcing their importance in risk assessment and pricing strategies. The findings confirm the suitability of the Gamma regression model for modeling skewed financial data in insurance and demonstrate the critical role of choosing the appropriate link function for accurate cost estimation.