Generalized Gamma Regression Model: Simulation Study and Its Application

A popular statistical tool that works particularly well in situations when the data distribution is positively skewed and asymmetrical is the generalized gamma (GG) distribution. In many real-world scenarios, multiple factors can influence different outcomes at the same time. The Generalized Gamma Regression (GGR) model, which is intended for answers that have a Generalized Gamma (GG) distribution, is presented in this article. The Berndt-Hall-Hall-Hausman (BHHH) algorithm is used to optimize the Maximum Likelihood Estimation (MLE) technique, which is used for parameter estimation in the GGR model. We employ the Maximum Likelihood Ratio Test (MLRT) and the Wald test for partial testing to evaluate the model's relevance. Thorough simulation validation shows that the GGR model is capable of accurately estimating parameters with little bias. To demonstrate its usefulness, we deployed the GGR model to a real-world case study. We specifically use it to examine the average years of schooling (AYS) in Central Java, Indonesia. In the final analysis, the current research emphasizes the advantages of applying the GGR model for generalized gamma distributed responses and demonstrates the model's resilience in parameter estimation.