Abstract
{ "background": "The adoption of advanced industrial machinery fleet systems, which integrate telematics and predictive maintenance, is critical for improving productivity and safety in heavy industries. However, quantifying and understanding the drivers of adoption rates in emerging economies remains methodologically challenging, with existing approaches often lacking the flexibility to model complex, multi-level industrial data.", "purpose and objectives": "This case study presents and evaluates a novel Bayesian hierarchical modelling framework designed to estimate and analyse the adoption rate of such systems within the South African industrial sector. The objective is to provide a robust methodological tool that accounts for heterogeneity across different industrial sub-sectors and company sizes.", "methodology": "A methodological case study was conducted, developing a Bayesian hierarchical logistic model. The model structure is given by $\\text{logit}(p{ij}) = \\alpha + \\alpha{\\text{sector}[i]} + \\beta X{ij}$, where $p{ij}$ is the probability of adoption for company $j$ in sector $i$, $\\alpha{\\text{sector}[i]}$ are sector-level random effects, and $X{ij}$ are firm-level covariates. Inference was performed using Hamiltonian Monte Carlo, with posterior credible intervals used to quantify uncertainty in adoption rate estimates.", "findings": "The application of the model to a proprietary industry dataset revealed substantial variation in adoption rates across sectors, with posterior estimates for the mining sector's intercept showing a 95% credible interval of [0.12, 0.45] on the log-odds scale. A key theme was the strong positive association between company asset value and adoption likelihood, with the model indicating this covariate had a high posterior probability (>0.98) of a positive effect.", "conclusion": "The Bayesian hierarchical model provides a statistically rigorous and interpretable framework for analysing technology adoption in industrial settings, effectively capturing sector-specific variability and firm-level influences.", "recommendations": "Industry analysts and policymakers should employ hierarchical modelling techniques to formulate targeted adoption strategies. Future research should integrate