Energy Forecasting in Commercial Buildings using Property Features and Natural Resources

Varghese, Basil (2024) Energy Forecasting in Commercial Buildings using Property Features and Natural Resources. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (517kB) | Preview

Abstract

This study aimed at using machine learning algorithms to forecast energy consumption in commercial buildings based on CBECS 2018 data. Principal component analysis brought the feature sets down from 1249 to 257 components while retaining 95% of the data variance. Four types of machine learning was used to the target variables with logarithmic transformation. The best hyper tuning done using RandomizedSearchCV, although applying 500 fits and determining the hyperparameter tuning with an MSE of 0.8954 and an MAPE of 5.67% was the XGBoost model applied with a learning rate of 0.1,a max depth of 3,and 300 no. of estimators out of 50 fits. Random Forest, tuned via RandomizedSearchCV across 50 fits, showed competitive results (MSE: 0.75, RMSE: 0.8651). Linear and Ridge Regression, two main models, gave baseline performance with an MSE of 0.9280 and 0.9314 respectively. This study also found that as far as renewable energy is concerned, natural gas is the most used source (70.6% of adoption across buildings), but electricity remains the most accepted traditional resource, Adoptions of renewable energy sources remain restricted to solar energy source, which was 4.4%. The study has shown that complex pattens of energy consumption are better represented by ensemble methodologies and that the penetration of renewable energy has high potential in existing commercial spaces.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Cosgrave, Noel UNSPECIFIED
Uncontrolled Keywords:	Commercial buildings; Machine learning algorithms; Principal Component Analysis; CBECS 2018
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science 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 Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	05 Sep 2025 13:37
Last Modified:	05 Sep 2025 13:37
URI:	https://norma.ncirl.ie/id/eprint/8831

Actions (login required)

View Item