A Comparative Analysis of Tree-Based, Deep Learning, and Ensemble Models for Building Energy Forecasting

Bujoreanu, Dan Alexandru (2025) A Comparative Analysis of Tree-Based, Deep Learning, and Ensemble Models for Building Energy Forecasting. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

Accurate short-term building energy load forecasting (STBELF) is essential for smart-grid operation, renewable energy sources flexible integration, and carbon-aware building management. This thesis develops a comprehensive STBELF framework, built around a four-stage pipeline that produces over 100 domain specific features, narrowed down to 35 core features/ predictors. On a heterogeneous, real-world dataset of 45 Norwegian public buildings, the study compares classical tree ensembles, deep learning architectures (LSTM, CNN-LSTM, TFT), and stacked ensembles.

Results show that, under rigorous feature engineering, tree-based models— Random Forest (MAE 3.30 kWh, R2 = 0.9819) and XGBoost (MAE 3.42 kWh, R2 = 0.9817)—significantly outperform deep learning models while training orders of magnitude faster. A stacking ensemble delivered competitive performance (MAE 3.58 kWh) but did not surpass the best single tree model. The findings support a feature-engineering-first methodology that prioritises data representation over model complexity as a pragmatic, robust strategy for real-world STBELF.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Onwuegbuche, Faithful Chiagoziem UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence H Social Sciences > HC Economic History and Conditions > Natural resources > Power resources > Energy consumption Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	28 May 2026 13:08
Last Modified:	28 May 2026 13:08
URI:	https://norma.ncirl.ie/id/eprint/9313

Actions (login required)

View Item