Green hydrogen produced via renewable-powered electrolysis is a key pathway for global decarbonization; however, its large-scale deployment remains limited by high costs and regional variability in renewable resources and infrastructure. To address the limitations of conventional deterministic techno-economic models, this study presents a machine learning-based framework that integrates spatial datasets, including renewable energy potential, electricity tariffs, and electrolyzer techno-economic parameters, across various regions of the world. A Random Forest model predicts the levelized cost of hydrogen with a test root mean square error of $0.28 per kg and an R² value of 0.94, significantly outperforming a simple linear regression model, which achieved an R² of 0.71. The main factors influencing cost are electricity price, availability of renewable energy, electrolyzer cost, and system efficiency. The model identifies the lowest-cost production areas, with hydrogen projected to cost between $1.8 and $2.5 per kg, in regions combining high solar and wind resources with stable infrastructure, such as North Africa, the Middle East, and Australia. This approach can inform policy, investment, and planning for cost-effective green hydrogen deployment.