{ "background": "Transport maintenance depots in Uganda face persistent challenges in resource allocation and operational planning, leading to suboptimal yield in parts refurbishment and vehicle availability. Existing management approaches often rely on reactive, historical averages rather than predictive analytics, limiting systemic improvement.", "purpose and objectives": "This article presents a methodological evaluation of a novel time-series forecasting model designed to measure and improve operational yield within these depot systems. The primary objective is to detail the model's architecture and validate its methodological rigour for forecasting key performance metrics.", "methodology": "The methodology integrates an autoregressive integrated moving average (ARIMA) framework with exogenous variables (ARIMAX) to account for seasonal maintenance cycles and resource input lags. The core model is specified as $Yt = \\mu + \\sum{i=1}^{p}\\phii Y{t-i} + \\sum{j=1}^{q}\\thetaj \\epsilon{t-j} + \\sum{k=1}^{m}\\betak X{t,k} + \\epsilont$, where $Yt$ is the yield metric. Model parameters were estimated using maximum likelihood, and robustness was assessed via rolling-origin forecast evaluations.", "findings": "The methodological evaluation demonstrates that the ARIMAX model significantly outperforms benchmark naive and simple exponential smoothing models, reducing the mean absolute percentage error (MAPE) by approximately 42% in out-of-sample testing. Forecast uncertainty, expressed as a 95% prediction interval, was found to be sensitive to the volatility of spare parts supply, a key exogenous variable.", "conclusion": "The proposed time-series forecasting model provides a statistically robust methodological framework for predicting depot yield, offering a substantial improvement over conventional planning tools. Its structured approach enables depot managers to transition from reactive to proactive maintenance scheduling.", "recommendations": "Implementation should be preceded by a depot-specific calibration phase to tailor exogenous variables. Training for engineering staff on interpreting forecast intervals is essential for operational adoption. Further research should explore integrating real-time inventory data