Abstract
<title>Abstract</title> <p>In today’s world of increasing energy demand and rising electricity costs, understanding and analyzing household electric power consumption has become essential. The study uses a multivariate time-series dataset to analyze household electric power consumption, predicting future patterns and identifying anomalies. It integrates Machine Learning (ML) models to predict consumption trends, enabling proactive energy management and sustainable living. The Individual Household Electric Power Consumption (IHEPC) dataset provides real-time data for correlations among electrical parameters, feature extraction, and model-building. In the current work, the effects of Decision Treee (DT), Random Forest (RF), Extreme Gradient Boosting (XGBoost), Gradient Boosting (GB), Light Gradient Boosting Machine (Light GBM) models on different characteristics are compared. The experiment shows that the effect of the RF regression model is better than other models in the experimental dataset. With the emergence of alternative methodologies over time, regression models’ accuracy limits in demand prediction have been stretched, as well as some of the restrictions that are commonly connected with their application. Model explainability was established with SHAP, which indicated Global_intensity had a clearly dominant impact on the prediction of electric power consumption. Methods based on ML algorithms are becoming increasingly popular, supports smart grid energy optimization and sustainable living.</p>