<![CDATA[Air quality is a crucial factor due to its significant impact on environmental sustainability and public health. One of the major pollutants affecting air quality is Nitrogen Monoxide (NO), especially during periods of increased human mobility such as Eid al-Fitr. Monitoring and predicting NO levels are essential for early mitigation efforts. This study aims to evaluate the performance of the Generalized Space-Time Autoregressive Integrated Moving Average (GSTARIMA) model with three types of spatial weighting schemes and compare it with other forecasting methods, namely ARIMA, VARIMA, and Support Vector Regression (SVR), in predicting NO concentrations in Surabaya for April 2024. The data used in this study consist of daily NO concentration measurements obtained from the Surabaya City Environment Agency’s monitoring stations located at SPKU Tandes, SPKU Wonorejo, and SPKU Kebonsari, covering the period from January 2023 to March 2024. The GSTARIMA model was selected for its capability to capture both spatial and temporal dependencies across monitoring locations. As an extension of the ARIMA model, GSTARIMA incorporates spatial weight matrices to model spatial heterogeneity. Parameter estimation was conducted using the Ordinary Least Squares (OLS) method. The results indicate that the GSTARIMA model with Inverse Distance Weighting (IDW) and order (3,1,0)₁ in the first spatial order yields the most accurate predictions, outperforming ARIMA, VARIMA, and SVR models. The model produced the lowest Symmetric Mean Absolute Percentage Error (sMAPE) of 0.93% and Root Mean Square Error (RMSE) of 5.32. A notable spike in NO concentrations was observed between April 23 and 25, 2024, coinciding with the post-Eid al-Fitr return flow, indicating a surge in population mobility.]]>
