Machine learning models show accuracy drop with limited residential energy data

A new study published on arXiv compares the effectiveness of various machine learning models for estimating residential energy consumption using limited input data. Researchers found that while models like CatBoost achieved high accuracy with full feature sets (R2=0.90 for ResStock, R2=0.73 for RECS), performance converged significantly when restricted to only ten easily accessible inputs (R2=0.61 for RECS, R2=0.62 for ResStock). However, for more homogeneous groups of homes, reduced-input models showed improved accuracy, suggesting that targeted modeling can be effective despite data limitations. AI

IMPACT Highlights the trade-offs between data availability and model accuracy in energy estimation, impacting the practical deployment of ML in this domain.

RANK_REASON Academic paper detailing a comparative analysis of machine learning models for energy estimation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• arXiv
• CatBoost
• Climate Zone 6A
• Hugging Face
• LightGBM
• Neural Networks
• random forest
• ResStock
• XGBoost

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →