Sampling Strategy Analysis of Machine Learning Models for Energy Consumption Prediction

With the development of the Internet of things (IoT), energy consumption of smart buildings has been widely concerned. The prediction of building energy consumption is of great significance for energy conservation and environmental protection as well as the construction of smart city. With the development of artificial intelligence, machine learning technology has been introduced to energy consumption prediction. In this study, multiple learning algorithms including Support Vector Regression (SVR), Artificial Neural Network (ANN), Random Forest (RF) are developed to perform energy consumption prediction. The most appropriate machine learning algorithm for energy consumption prediction has been investigated and found to be the random forest algorithm. Based on the developed machine learning models, studies on the sampling strategy for energy consumption prediction have been conducted. It is found that the variance of data has a significant effect on the prediction accuracy, and a better prediction result can be achieved by increasing the sampling density over the data with high variance. This result can be used to optimize the machine learning algorithm for building energy consumption prediction and improve the computational efficiency.