The European Union Emissions Trading System (EU ETS) stands as the pioneering global emissions trading system, with carbon trading prices, especially EU carbon allowance prices, assuming a pivotal role in achieving carbon neutrality and advancing low-carbon initiatives. The fluctuating EU carbon price has emerged as a vital reference point for shaping policies and informing decisions among firms, policymakers, and investors. Yet, the precise forecasting of carbon prices holds significant implications for national energy resilience and sustainability policies. Machine learning techniques offer enhanced accuracy, hence heightened ability to identify irregularities and increased adaptability and versatility. In this study, we utilize the boosted regression trees (BRT) learning technique to predict the EU carbon emissions trading price and evaluate the influence of predictor variables on the system's response. Our analysis incorporates key determinants of ETS prices, encompassing oil, natural gas, coal, and European stock market. Additionally, we consider financial and economic factors, such as global uncertainties, geopolitical risks, pandemic-related concerns, bond market uncertainty, and financial stress factors. The findings demonstrate the robust predictive capability of BRT in forecasting EU carbon prices. The STOXX Europe 600 index and pandemic-related uncertainty emerge as primary factors influencing carbon emissions. Policymakers should integrate economic policies with climate goals to enhance investor confidence in carbon markets. They should also bolster crisis response and risk management strategies to mitigate the impact of global health crises and other shocks on carbon pricing.