<![CDATA[Reliable revenue forecasting is critical for ensuring the financial sustainability of urban sanitation infrastructure, particularly in publicly managed fecal sludge treatment systems where demand fluctuates and operational planning depends on daily service variability. However, revenue patterns in such systems are typically nonlinear, volatile, and influenced by interrelated operational factors, limiting the effectiveness of conventional linear forecasting approaches. This study develops a data-driven predictive framework using Support Vector Regression (SVR) to model daily retribution revenue at the Keputih Fecal Sludge Treatment Plant (IPLT Keputih), Surabaya. The dataset comprises 1,213 daily observations from January 2022 to April 2025, incorporating total transactions and total sludge volume as predictor variables and total revenue as the response variable. Three kernel configurations—Linear, Polynomial, and Radial Basis Function (RBF)—were systematically evaluated following Min–Max normalization and chronological training–testing separation. Model performance was assessed using Mean Absolute Percentage Error (MAPE). The results demonstrate that the SVR model with the RBF kernel achieved the highest predictive accuracy, yielding a MAPE of 17.17%, outperforming the Linear and Polynomial kernels in capturing nonlinear revenue dynamics. Forecast projections further reveal cyclical seasonal tendencies with direct implications for operational scheduling and short-term budget allocation. By integrating machine learning–based forecasting into public sanitation revenue modeling, this study contributes to advancing data-driven financial planning strategies for sustainable urban service management.]]>
