MACHINE LEARNING-BASED MODELING FRAMEWORK FOR IMPROVING ROMANIAN RESILIENCE STRATEGY TO GREENHOUSE GAS EMISSIONS IN RELATION TO VISEGRAD GROUP

The present research reveals the difference between Romania and V4 in terms of the Greenhouse Gas Emissions Strategy and establishes a machine learning (ML) -based modeling framework for improving the ability to reach zero GHG by the mid-21st century. The ML tree-based algorithms, based on dual dimension environmental-economic nexus, revealed that net greenhouse gas emissions (NGHGE) are mostly conditioned by greenhouse gases from agriculture (GHGA), a fact valid both in the case of Romania (feature importance -FI = 0.41) and V4 (FI = 0.86). However, for V4, the 2nd important predictor is identified as greenhouse gases from waste management (FI = 0.26), while in the case of Romania, the national expenditure on environmental protection has a limited impact (FI =0.02) on NGHGE. Both integrated models have good prediction accuracy (Rsq. 0.70, RMSE 0.53 for the model associated with the Romania database and Rsq. 0.76, RMSE 0.47 for the V4 model). It can be concluded that in terms of integrated GHG emissions management strategy, Romania can merge with V4 to increase the environmental efficiency towards achieving the EU environmental goals.