Forecasts of household energy consumption as the main consumption sources in the field of energy economics through a machine learning model
Forecasts of household energy consumption as the main consumption sources in the field of energy economics through a machine learning model
nasim omodi1
1) Master s student in Energy Economics, Ilam University
Publication :
3rd International Congress on Science, engineering & New Technologies(secongress.com)
Abstract :
Optimal energy management in the household sector, as one of the most important consumption sources in the energy economy, plays a key role in achieving sustainability, reducing carbon emissions, conserving natural resources, and saving money. However, traditional methods of predicting energy consumption often lack sufficient accuracy, which highlights the need for advanced approaches. In this regard, artificial intelligence (AI) has been proposed as a powerful tool for modeling and predicting energy consumption; however, AI-based models usually lack sufficient transparency and interpretability due to their black box structure. To address this challenge, explainable artificial intelligence (XAI) frameworks have been developed to enable better analysis and understanding of machine learning model decisions. In this study, the goal is to achieve accurate prediction of household energy consumption using machine learning models and compare their performance based on various evaluation criteria, including coefficient of determination (R2), root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE). The superior model is identified through testing on unseen data and then analyzed using two XAI frameworks, including local model-independent explanation (LIME) and incremental Shapley quantities (SHAP). These analyses reveal the effective features in energy consumption prediction and provide practical insights into their significance. The results emphasize the importance of past consumption patterns and the role of low consumption values in accurate energy prediction. Also, this study shows that the use of XAI frameworks can greatly contribute to the development of more transparent, consistent, and reliable forecasting models.
Keywords :
home energy consumption
machine learning
explainable artificial intelligence (xai)
energy economics
model evaluation