Analysis of Energy Conservation Awareness and Behavior through Big Data and Machine Learning

Abstract

This study derives a machine learning model to predict household energy conservation awareness using electric power data obtained from over 200 smart houses in Kitakyushu City, along with survey results. We employ the Gradient Boosting Decision Tree (LightGBM) as a machine learning model and reveal important features contributing to the prediction by ranking significant feature variables. This research presents a challenging task of predicting psychological variables such as energy conservation awareness and behavior from physical data like electricity usage.