Predicting CO2 emissions in Morocco: exploring the use of ridge regression with data preprocessing and feature impact analysis.

This study applies ridge regression (RR) to predict CO2 emissions in Morocco and compares its performance with LASSO regression, ElasticNet regression, and linear regression. The dataset spans 1965-2021 and includes ten variables representing energy, economic, and demographic factors. The RR model, trained on 1965-2009 data and validated on 2010-2021, achieves the highest accuracy (R2 = 0.994), outperforming the other regression models. Feature importance analysis reveals that oil consumption is the dominant driver of emissions, accounting for 129% of the standardized effect, while coal consumption contributes around 9%. In contrast, renewables and hydro generation show negative effects, highlighting their mitigating potential. These percentages represent the relative weight of each factor in shaping CO2 emissions. Policy scenario testing indicates that reaching Morocco's 2030 renewable energy target (52% share) could lower emissions by about 0.67 Mt annually, with deeper reductions under more ambitious renewable expansion. Overall, this research demonstrates the reliability of ridge regression for environmental modeling and provides policymakers with quantitative insights into Morocco's energy-emissions nexus, reinforcing the strategic value of renewable deployment for climate mitigation.