Energy consumption prediction of appliances using machine learning and multi-objective binary grey wolf optimization for feature selection

Abstract

Prediction of the energy consumed by household appliances is a challenging research topic owing to a transition toward the Internet of Everything. Although classical machine learning algorithms have evolved significantly in recent years owing to the advancements in big data and the Internet of Things, several challenges remain to be addressed. Thus, obtaining highly accurate predictions using as few computational resources as possible is the primary research goal of many studies. In this study, the energy consumptions of appliances were predicted using a method based on multi-objective binary grey wolf optimization, wherein the random forest, extra trees, decision tree, and K-nearest neighbor regression algorithms were employed. The two objectives of the present study were the maximization of the prediction performance of the algorithms and minimization of the number of selected features. The results were ranked using multi-objective optimization on the basis of ratio analysis. The proposed method was tested on the appliances energy prediction dataset publicly available at the UCI Machine Learning Repository, and its results were compared with those obtained using other similar methods.