"background": "Industrial machinery fleets in Kenya face persistent challenges in operational yield, with existing maintenance and scheduling systems often failing to optimise performance. A lack of robust, data-driven forecasting tools tailored to local operational conditions hinders systematic improvement. ", "purpose and objectives": "This study aimed to methodologically evaluate current fleet management systems and to develop a bespoke time-series forecasting model for predicting and improving machinery yield in a Kenyan industrial context. ", "methodology": "A hybrid methodology was employed, integrating a diagnostic evaluation of fleet management practices with the development of an Autoregressive Integrated Moving Average with exogenous variables (ARIMAX) model. The model, specified as Yt = \ + =1^{p\ Yt-i + =1^q\ -j + =1^r\ Xt, k + \ₜ, was trained and validated using high-frequency operational data from a fleet of earth-moving equipment. ", "findings": "The diagnostic evaluation revealed that 68% of fleets relied on reactive maintenance strategies. The ARIMAX model achieved a statistically significant forecast accuracy, with a mean absolute percentage error (MAPE) of 7. 3% (95% CI: 6. 8, 7. 9) for weekly yield, outperforming benchmark models. Predictive maintenance scheduling informed by the model was shown to be the primary driver of potential yield gains. ", "conclusion": "The developed forecasting model provides a quantitatively robust tool for yield prediction, demonstrating that a shift from reactive to predictive management is both feasible and advantageous for industrial machinery operations in the studied region. ", "recommendations": "Fleet managers should adopt predictive, data-driven models for maintenance scheduling. Further research should focus on integrating real-time sensor data to enhance model granularity and adaptability across different machinery types. ", "key words": "fleet management, predictive maintenance, ARIMAX, operational yield,
Ochieng et al. (2004) studied this question.