Analyzing energy consumption trends and environmental influences: A time-series study on temperature, renewables, and demand correlations

Abstract

This study investigates energy consumption trends and environmental influences by analyzing time-series data to explore the correlation between temperature, humidity, renewable energy contributions, and energy demand. The research focuses on developing an advanced hybrid machine learning model using LightGBM, XGBoost, LSTM, and SHAP to enhance the accuracy and interpretability of energy consumption predictions. Using data from January 2022 to January 2025 across residential, commercial, and industrial buildings, the study examines the impact of temperature fluctuations, humidity, and renewable energy integration on energy consumption. Temperature dependency is further explored in the study, where it is shown that energy consumption is directly influenced by temperature, with energy use at 20 °C being 2000 kWh, increasing to 3200 kWh at 30 °C (on an annual basis), further confirming the shaped dependency with increased cooling demands during warmer months. Additionally, energy consumption varies significantly across building types, with industrial buildings showing higher and more stable energy demands than residential and commercial buildings. Results indicate that XGBoost provides the best predictive performance, with an RMSE of 118.24 and an R² score of 0.9871, followed by LSTM with an RMSE of 135.86 and an R² score of 0.9752, and Linear Regression with RMSE of 187.76 and an R² score of 0.9672. The hybrid model effectively predicts energy consumption and offers valuable insights into how environmental factors influence energy demands across different building types. The findings contribute to optimizing energy management strategies, improving innovative grid systems, and promoting sustainable building operations while highlighting the role of renewable energy in shaping energy consumption patterns.