Predicting Site Energy Usage Intensity Using Machine Learning Models.

Area of Science:

• Environmental Science
• Data Science
• Machine Learning

Background:

• Climate change, driven by human activities, necessitates energy consumption evaluation.
• Predicting energy usage is crucial for identifying factors and reducing waste.
• Machine learning is increasingly applied to energy-related research.

Purpose of the Study:

• To predict building energy consumption using machine learning.
• To identify key factors influencing energy usage in buildings.
• To provide actionable insights for reducing energy waste.

Main Methods:

• Utilized the Women in Data Science (WiDS) Datathon 2022 dataset, including building characteristics and sensor data.
• Applied and compared four machine learning models: Random Forest (RF), Gradient Boost Decision Tree (GBDT), Support Vector Regressor (SVR), and Decision Tree for Regression (DT).
• Evaluated model performance using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE).

Main Results:

• The study successfully demonstrated the effectiveness of machine learning in predicting building energy consumption.
• Identified hidden patterns and insights within the dataset related to energy usage.
• The most performant model was selected based on RMSE and MAE evaluation metrics.

Conclusions:

• The proposed machine learning approach robustly captures patterns for accurate energy usage prediction.
• Effective energy prediction can guide strategies for reducing consumption and mitigating climate change.
• This research highlights the potential of data science in addressing environmental challenges.