Parameter estimation and hypothesis testing of geographically weighted generalized log gamma regression model

In area-based research, there are often spatial effects that make global modelling less accurate. This is because each location has different characteristics. Therefore, the Generalized Log Gamma Regression model (GLGR) becomes less appropriate to apply. In this research, a Geographically Weighted Generalized Log Gamma Regression model (GWGLGR) is developed as a solution to the problem, that generates distinct parameters for every observation site. This study aims to estimate the parameters and obtain test statistics of GWGLGR model. The results show that the estimator of GWGLGR model can be obtained by using the Maximum Likelihood Estimation (MLE) and the BHHH algorithm, also obtains a covariance matrix estimator from its parameter vector estimator. The test statistics to test the GWGLGR model hypothesis simultaneously obtained using the Maximum Likelihood Ratio Test (MLRT), the GLGR model similarity test with GWGLGR using the F test while partially, the Z test is obtained using the central limit theorem.