<![CDATA[The degradation of river water quality in East Kalimantan has become an environmental issue related to the Sustainable Development Goals (SDGs), particularly Goal 6 concerning clean water and sanitation. This study aims to model Chemical Oxygen Demand (COD) as an indicator of river water quality using the Fixed Effect Model (FEM) and Geographically Weighted Panel Regression (GWPR) approaches, as well as to compare the performance of both models in representing spatial and temporal heterogeneity. The data used in this study consist of panel data from 31 observation locations over four semesters, with predictor variables including pH, Total Suspended Solid (TSS), Fecal Coliform, Total Dissolved Solid (TDS), and ammonia. FEM estimation was conducted using the within estimator, while the GWPR model employed an adaptive bisquare kernel weighting function. The results show that TSS and ammonia significantly affect COD in the FEM model. In the GWPR model, the effects of predictor variables vary across observation locations, indicating spatial heterogeneity in the factors influencing COD. TSS was identified as the most dominant variable, being significant at 26 locations, followed by ammonia and TDS, which were significant at 18 and 15 locations, respectively. Furthermore, the GWPR model produced a lower RMSE value (1.3777) than the FEM model (2.182011). These findings indicate that GWPR performs better in capturing spatial heterogeneity and temporal information in river water quality and provides more detailed information for supporting location-specific water quality management in East Kalimantan.]]>
