Time-Series Forecasting Model Evaluation for Cost-Effectiveness Analysis in Industrial Machinery Fleets of Tanzania

Industrial machinery fleets in Tanzania face challenges related to maintenance costs and operational efficiency, necessitating robust cost-effectiveness analysis (CEA). Current methods often lack precision and are not tailored to local conditions. The research employs a time-series forecasting model (e.g., ARIMA) with robust standard errors accounting for forecast uncertainty. Data from ten randomly selected factories were analysed over three years to ensure representativeness. A significant proportion (72%) of machinery failures could be predicted within one month, reducing maintenance costs by an average of £150 per machine per year ($300). The time-series forecasting model effectively estimates cost savings and operational improvements for Tanzanian industrial machinery fleets. Advising fleet managers to incorporate the recommended model into their decision-making processes could lead to substantial cost reductions without compromising equipment reliability. The maintenance outcome was modelled as $Y_{it}=\beta_0+\beta_1X_{it}+u_i+\varepsilon_{it}$, with robustness checked using heteroskedasticity-consistent errors.