Penerapan Metode Geographically Weighted Logistic Regression Untuk Memodelkan Pencemaran Air Sungai Mahakam Berdasarkan Data Dissolved Oxygen

Abstract

The Geographically Weighted Logistic Regression (GWLR) model is a local model of logistic regression applied to spatial heterogenity data. Parameter estimation of the GWLR model is conducted at each observation location using spatial weighting. The aim of this research is to obtain the GWLR model on the Dissolved Oxygen (DO) data of Mahakam River in 2022, and to identify the factors affecting the probability of Mahakam River water is polluted. The research data is secondary data obtained from Environmental Department of East Kalimantan Province. Spatial weight is calculated using the adaptive bisquare weighting function, and the optimal bandwidth is determined using the Generalized Cross Validation (GCV) criterion. Parameter estimation method is Maximum Likelihood Estimation (MLE), and Maximum Likelihood (ML) estimator was obtained using the iterative Newton-Raphson method. Based on the result of the GWLR model parameter testing, it was concluded that locally influential factors on the probability of Mahakam River water pollution are nitrate concentration and iron concentration, and globally influential factor is nitrate concentration.