Upaya Pencegahan Pencemaran Air Sungai Mahakam melalui Pemodelan Geographically Weighted Logistic Regression pada Data BOD

Abstract

Since the early years, Mahakam River has important roles in supporting human needs in East Kalimantan province. Activities around Mahakam watershed such as restaurants, fishery, and industries were in the potential of generating waste around the flow area. The waste consisted of domestic and nondomestic waste. The waste was a threat to the Mahakam River water quality. Water pollution around the Mahakam River was a threat to public health, and therefore, there’s a need for precaution. One of the precautions is to give the public information regarding the factors that influence the chances of polluted water in the Mahakam River increased through logistic regression modeling. One way to detect water pollution is to indicate by using Biochemical Oxygen Demand (BOD). BOD data was suspected spatial, therefore the appropriate statistical modeling is Geographically Weighted Logistic Regression (GWLR). GWLR is a regression model that developed from a logistic regression in which parameter estimation is done locally at every observation location. The purpose of the research is to determine the GWLR model on the BOD data of Mahakam River and to find out the factors that influence water pollution at 27 observation points along with the Mahakam River flow. The parameter estimation method is the Maximum Likelihood Estimation (MLE). The spatial weighting is calculated by using the Adaptive Bisquare weighting function and the optimum bandwidth is determined by using Generalized Cross-Validation (GCV) criteria. Research shows that the closed-form of the Maximum Likelihood estimator can’t be obtained analytically and the approximation is obtained by using Newton-Raphson (N-R) iterative method. Based on parameter testing of the GWLR model result, it was concluded that the factors were influences the probability of Mahakam River water were polluted based on the BOD indicator was locally and different in each 27 observation locations. The factors that influence locally were water temperature, acidity, Total Dissolved Solids (TDS), ammonia concentration, and water debit, meanwhile, the factors which influence globally were acidity and TDS.