Comparative Modeling of Green Hydrogen Development in Costa Rica and the UK: A Machine Learning-Driven Policy and Investment Forecasting Approach

Green hydrogen is gaining prominence as a strategic option for decarbonizing sectors that are difficult to electrify, yet its economic potential differs across regions. This study develops a comparative modeling framework enhanced by machine learning to project both the Levelized Cost of Hydrogen (LCOH) and Net Present Value (NPV) for hydrogen projects in Costa Rica and the United Kingdom (UK). By combining geospatial energy resource mapping, techno-economic modeling, and advanced tools such as Random Forest algorithms, SHAP interpretability analysis, and Monte Carlo simulations, the research investigates how hydrogen production costs shift under varying policy and infrastructure scenarios. Findings show that Costa Rica’s LCOH ranges from $3.4 to $5.1 per kilogram, influenced by renewable energy type and system scale, while the UK benefits from financial incentives that help offset its higher baseline costs. Across both countries, the key determinants of cost were electricity pricing, capital expenditure, and electrolysis efficiency. The use of machine learning significantly improved prediction accuracy and allowed for deeper exploration of policy sensitivities. The outcomes not only align with each nation’s hydrogen strategy but also point to actionable avenues for international cooperation, including joint technology development, pilot programs, and hybrid financing models. This approach demonstrates how data-driven analysis can support more equitable and resilient transitions to zero-carbon energy systems, particularly when supported by interpretable artificial intelligence methods.