Time-Series Forecasting Model Evaluation for System Reliability in Ethiopian Manufacturing Plants

This study evaluates system reliability in manufacturing plants of Ethiopia, focusing on time-series forecasting models to enhance understanding and predictability. A mixed-method approach was employed, integrating both quantitative data from historical plant operation records and qualitative insights from expert interviews to evaluate the time-series forecasting models. The application of ARIMA (Autoregressive Integrated Moving Average) models demonstrated a significant improvement in predicting system failures with an RMSE error rate of less than 5% compared to baseline methods, indicating enhanced reliability measures. The results suggest that time-series forecasting models can be effectively utilised for improving the operational efficiency and reliability of manufacturing systems in Ethiopia, particularly through ARIMA model application. Manufacturing plants should consider adopting time-series forecasting models as a proactive strategy to enhance system performance and minimise downtime. Future research could explore incorporating machine learning techniques into these models. time-series forecasting, system reliability, Ethiopian manufacturing, ARIMA model The maintenance outcome was modelled as $Y_{it}=\beta_0+\beta_1X_{it}+u_i+\varepsilon_{it}$, with robustness checked using heteroskedasticity-consistent errors.