Geographically and temporally weighted multivariate generalised gamma regression for modelling three educational indicators in Central Java, Indonesia

Goal 4 of the United Nations Sustainable Development Goals, which focuses on quality education, is relevant in the context of Central Java.With varying achievements concerning educational indicators across districts/cities and skewed-positive distribution, Central Java must address educational challenges and improve the quality of students' learning experiences.While many regression-based models have been introduced, no model has captured spatio-temporal variation for skewed-positive distributions.This study proposes a geographically and temporally weighted multivariate generalised gamma regression (GTWMGGR) model and applies it to model three educational indicators in Central Java, encompassing the mean years of schooling (MYS), school enrolment rate (SER) and gross enrolment rate (GER).We develop parameter estimation using the maximum likelihood method combined with Berndt-Hall-Hall-Hausman algorithm.In addition, we derive the hypothesis test procedure using the maximum likelihood ratio method.The results indicate that the GTWMGGR model parameters are significant at any given point., revealing eleven groups for MYS and GER indicators and nine groups for SER indicators.By identifying the specific educational contexts for each group, our study highlights the importance of developing tailored strategies and interventions to improve educational outcomes in Central Java.Geographically and temporally weighted multivariate generalised gamma regression for modelling three educational indicators in Central Java, Indonesia 259