<![CDATA[Accurate prediction of household energy consumption plays a vital role in smart grid management, demand-side response, and energy efficiency planning. This study presents a comparative analysis of four ensemble machine learning models—Random Forest, XGBoost, LightGBM, and CatBoost—for predicting household active power consumption. The publicly available Household Electric Power Consumption dataset from the UCI repository, comprising over two million measurements, was used for experimentation. Three feature scenarios were designed to evaluate model performance under different input conditions: using all features, excluding current intensity, and excluding both intensity and reactive power. Results show that Random Forest achieved the highest predictive accuracy with an R² of 0.913 in the reduced-feature scenario, outperforming gradient boosting alternatives. Statistical validation using the Wilcoxon signed-rank test was used to assess the effect of feature exclusion across the three scenarios. The study demonstrates that reliable energy prediction remains achievable using sub-metering, voltage, and temporal features, offering a practical baseline for smart meter applications.]]>
Copyrights © 2026
