Process-control systems in South Africa are employed to optimise yield in agricultural settings. However, their effectiveness varies widely and lacks a standardised method for forecasting yield improvements. We developed a novel time-series forecasting model to predict yield outcomes. The methodology involved collecting historical data from multiple agricultural sites across South Africa, applying advanced statistical techniques such as ARIMA (AutoRegressive Integrated Moving Average) for analysis. Our findings indicate that the ARIMA model significantly improved forecast accuracy by reducing prediction errors by an average of 15% compared to existing methods. This precision is crucial for resource allocation and policy formulation in agricultural sectors. The robustness of our time-series forecasting model validates its utility in enhancing yield improvement predictions, offering a methodological framework that can be replicated across diverse agricultural settings in South Africa. Aimed at policymakers, we recommend the adoption of this forecasting tool to inform strategic decisions regarding investment and resource allocation for maximum efficiency in agricultural processes. The maintenance outcome was modelled as $Y_{it}=\beta_0+\beta_1X_{it}+u_i+\varepsilon_{it}$, with robustness checked using heteroskedasticity-consistent errors.