Analisis Faktor-Faktor Yang Berpengaruh Terhadap Pencemaran Air Sungai Mahakam Menggunakan Pemodelan Geographically Weighted Logistic Regression Pada Data Dissolved Oxygen

Abstract

Geographically Weighted Logistic Regression (GWLR) model is a local model of the logistic regression model applied to spatial data. Parameter estimation is performed at each observation location using spatial weighting. The spatial weighting is calculated by using an adaptive tricube function and bandwidth optimum was obtained based on Generalized Cross Validation (GCV) criteria. The purpose of this study was to obtain a GWLR model on the water pollution indicator Dissolve Oxygen (DO) in Mahakam River in East Kalimantan Province and to find factors affecting the probability of the Mahakam River water was not polluted based on DO indicator. The research data is secondary obtained from Environmental Department of East Kalimantan.  The parameter estimation method was Maximum Likelihood Estimation (MLE). The research result showed that the closed form of ML estimator could not be found analytically and it can be approximed by using Newton-Raphson iterative methods. Based on the result of partial hypothesis test, the factors influencing the probability of the Mahakam River water was not polluted is different for every observation location. They were phosphate consentration, total dissolved solid and nitrite consentration. The factor influencing globally was total dissolved solid.