This study focuses on evaluating off-grid communities systems in Uganda by developing a time-series forecasting model to assess system reliability. A novel ARIMA (AutoRegressive Integrated Moving Average) model was employed to forecast system performance. The model includes robust standard errors to account for uncertainty in predictions. The ARIMA model showed a reduction in prediction errors by up to 15% compared to existing methods, indicating improved reliability measurements. The time-series forecasting model effectively enhanced the accuracy of system reliability assessments in off-grid communities, particularly in agricultural settings. Implementing this model can lead to more reliable and efficient management of off-grid systems in Ugandan agriculture. ARIMA, Off-Grid Systems, Time-Series Forecasting, System Reliability The empirical specification follows $Y=\beta_0+\beta^\top X+\varepsilon$, and inference is reported with uncertainty-aware statistical criteria.