Water treatment facilities in Ethiopia have faced challenges in maintaining consistent operational reliability due to varying water quality inputs and fluctuating demand. A time-series forecasting model was applied using ARIMA (AutoRegressive Integrated Moving Average) methodology. Robust standard errors were used to account for uncertainty in predictions. The model identified a significant positive correlation between input water quality and system efficiency, with an R² value of 0.85 indicating substantial explanatory power. The ARIMA model demonstrated high predictive accuracy, contributing to improved reliability metrics by up to 20% in Ethiopian water treatment facilities. Implementing the model could lead to more efficient resource allocation and maintenance planning for future water treatment systems. water quality forecasting, time-series analysis, ARIMA, Ethiopian water treatment The maintenance outcome was modelled as $Y_{it}=\beta_0+\beta_1X_{it}+u_i+\varepsilon_{it}$, with robustness checked using heteroskedasticity-consistent errors.