Machine learning models for predicting rural residential carbon emissions and optimising spatial forms.

Area of Science:

• Environmental Science
• Urban Planning
• Sustainable Development

Background:

• Global warming, driven by carbon emissions, impacts ecosystems and populations.
• Spatial form is key to urban energy efficiency but understudied in rural areas.

Purpose of the Study:

• Investigate carbon emissions and spatial form in rural residential areas.
• Predict carbon emissions and optimize spatial form using advanced modeling techniques.

Main Methods:

• Employed Random Forest, XGBoost, and BP Neural Network models.
• Analyzed spatial form factors like floor area ratio, number of floors, and building orientation.
• Predicted carbon emissions and optimized spatial form for reduced environmental impact.

Main Results:

• Spatial form factors strongly correlate with carbon emissions.
• XGBoost model achieved superior prediction accuracy and generalization.
• Optimized spatial form led to over 10% reduction in carbon emissions.

Conclusions:

• Spatial form optimization is crucial for low-carbon rural development.
• Regulating floor area ratio and building shape coefficient are effective strategies.
• Findings support a green transition in rural areas through scientific planning.