Volume 42: Energy Transitions toward Carbon Neutrality: Part V

Unraveling Energy Consumption patterns: Insights through Data Analysis and Predictive Modeling Neda Maleki, Xianwei Xie, Arslan Musaddiq, Tobias Olsson, David Mozart, Fredrik Ahlgren

Abstract

Most of the utility meters in Sweden are connected using the Internet of Things (IoT) technology. This opens new possibilities for understanding society’s energy consumption dynamics and making citizens aware of their power consumption usage. In this study, we investigate the patterns of electricity consumption using machine learning methods. We collected metered data from Kalmar Energi company, the electrical grid for Kalmar city in Sweden. In addition, we collected the Kalmar weather and electricity price data from the Swedish Meteorological and Hydrological Institute (SMHI) and Nordpool, the European leading power market, respectively. We comprehensively analyze the electricity consumption data to assess the changes in overall electricity demand during the year 2021 in the city of Kalmar. This information can be of significant benefit to other regions seeking to improve their sustainability and energy consumption practices. For analysis and energy consumption prediction, we utilize two forecasting models, i.e., Random Forest (RF) and XGBoost. RF model results show a high level of accuracy with the achieved R-squared (R2) value of 0.91 compared to XGBoost value of 0.87.

Keywords energy consumption, machine learning, energy forecasting, Internet of Things