The Application of Machine Learning and Deep Learning Techniques for Global Energy Utilization Projection for Ecologically Responsible Energy Management

Accurately estimating future energy consumption is critical as the world seeks

alternatives to fossil fuels amidst rising energy demands. The research employs

various prediction models for global energy prediction with GDP analysis in

energy consumption context. These models include Regression models that are

Linear, Polynomial, Bayesian, Tree, Extreme Gradient Boosting, K Nearest

Neighbour, Stacked Model, Random Forest (RF), also Long Short-Term Memory

(LSTM) and Convolution Neural Networks (CNN) methods. Models are employed

to enhance global energy consumption modelling, analysing their adaptability to

varying weather and social conditions. A comparative investigation shows that RF

performs better than other Regression models. LSTM models perform better than

RF in predicting the primary energy consumption per capita and GDP growth,

with the lowest MSE value of 0.002 with comparatively higher time and processing

complexity. However, RF outperforms in predicting renewable energy share,

access to clean cooking fuel, CO2 emission and GDP per capita analysis. The

study's novelty lies in its comprehensive evaluation of machine learning and deep

learning methods across multiple geographic and temporal energy consumption

patterns, emphasizing the superiority of advanced techniques in accurately

modelling global energy usage.