This study evaluates time-series forecasting models in process-control systems within an Ethiopian agricultural context. A comparative analysis of ARIMA (AutoRegressive Integrated Moving Average) and LSTM (Long Short-Term Memory) neural network models was conducted using historical agricultural data from Ethiopian seed production facilities. Model performance metrics included mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination ($R^2$). The ARIMA model outperformed LSTM in terms of MAE, with a reduction in forecast errors by approximately 15%. LSTM models showed higher $R^2$ values, indicating better predictive accuracy. Both models were robust to the data's autocorrelation and seasonality. Further research should explore ensemble methods combining ARIMA and LSTM for enhanced forecasting in seed production systems. Agriculture, Ethiopia, Time-series Forecasting, ARIMA, LSTM