On the use of different link functions in gamma-pareto regression model: Simulation and application

The Gamma-Pareto regression model (G-PRM) is appropriate when the response variable follows Gamma-Pareto distribution (G-PD) is used as the generalized linear model (GLM).For the estimation of the G-PRM the Iterative weighted least squares (IWLS) method is used with a specific link function.In this study, we consider G-PRM under different link functions.However, the researchers do not pay much attention to the selection of suitable link functions.In the context of the G-PRM, three link functions are used, which allow for investigating how inverse, identity, and log link functions perform.A Monte Carlo simulation and a demonstration with real data were used to compare the performance of different link functions in G-PRM.The sum squared residual (SSR), mean squared error (MSE) and average mean squared error (AMSE) are used as the evaluation criterion for suitable link function.Both the simulation and real data findings demonstrate that the G-PRM with the identity link function provides efficient results having minimum SSR, MSE and AMSE.Advantages of the paper, choice of suitable link function always significantly impacts the model`s performance.