"background": "Industrial machinery fleet management in developing economies is critical for infrastructure development but is often hampered by sparse, heterogeneous data and high operational uncertainty. Traditional reliability models frequently fail to capture the complex, multi-level risk factors inherent in such contexts, leading to suboptimal maintenance and safety interventions. ", "purpose and objectives": "This article presents a novel Bayesian hierarchical modelling framework designed to quantify and reduce operational risks within industrial machinery fleets. The primary objective is to provide a robust methodological tool for integrating disparate data sources to estimate failure probabilities and identify dominant risk drivers. ", "methodology": "A three-level hierarchical model is developed, where machinery-specific failure events are modelled conditionally on equipment-level parameters (e. g. , age, maintenance history), which themselves are drawn from fleet-wide distributions. The core structure is defined by y{ij \ (pij), \\; (pij) = + \ Xij, \\; \ (\\, \\), with inference performed via Hamiltonian Monte Carlo. Prior distributions are informed by expert elicitation from local engineers. ", "findings": "Application to a case study fleet demonstrates the model's capacity to synthesise incomplete records. A key finding is the quantification of uncertainty, with the posterior distribution for the baseline failure odds ratio of a critical component showing a 95% credible interval of 1. 4, 3. 2. The analysis identifies inadequate preventative maintenance scheduling as the dominant systemic risk factor, outweighing the influence of equipment age alone. ", "conclusion": "The proposed Bayesian hierarchical model provides a statistically rigorous and context-adapted methodology for risk assessment in data-scarce environments. It formally incorporates uncertainty, enabling more informed, evidence-based decision-making for fleet managers. ", "recommendations": "Fleet operators should adopt this modelling framework to move from reactive to predictive maintenance strategies. It is recommended that
Uwimana et al. (Thu,) studied this question.