Machine learning-based predictive modelling of renewable energy adoption in developing countries

This study explores global renewable energy trends in alignment with the 2030 Sustainable Development Goals. Employing and fine-tuning the ExtraTreesRegressor, models were developed to predict adoption levels of electricity from solar, wind, hydro, and biomass sources. Strategic random search parameters were used to optimize the ExtraTreesRegressor. Evaluation based on Mean Square Error (MSE) and R-squared (R 2 ) scores revealed that the ExtraTreesRegressor, outperformed other state-of-the-art regression models. Notably, the solar model exhibited commendable performance in test set evaluation (MSE: 0.4450, R 2 : 0.9849) and cross-validation (MSE: 4.3279, R 2 : 0.9079). Similarly, the wind model showed robust outcomes in both test set evaluation (MSE: 1.2233, R 2 : 0.9969) and cross-validation (MSE: 5.3136, R 2 : 0.9846). However, the hydro model faced nuanced challenges with test set evaluation (MSE: 33.3474, R 2 : 0.9960) and cross-validation (MSE: 20.4235, R2: 0.9961). The biomass model achieved notable results in test set evaluation (MSE: 0.3196, R 2 : 0.9960) and cross-validation (MSE: 0.5943, R 2 : 0.9901). Based on the findings from this study, GDP, non-renewable electricity consumption, and population size have been identified as key drivers of renewable energy adoption. Insights from this research will contribute to a deeper understanding of the intricate dynamics influencing renewable energy landscapes in developing countries.