A data-driven and risk-based decision support system for the retrofit performance of buildings

Date

2025-8-25

Author

Bıyık, Hatice

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

329
views

0
downloads

Cite This

Retrofits are conducted to achieve higher energy efficiency and lower carbon emissions from buildings. Retrofits are generally based on energy, rarely on costs, and carbon emissions. However, retrofit performance gaps may occur between the designed and applied retrofits. Since risks and uncertainties may lead to changes in retrofit processes, they can be the main reasons for these gaps. Rapid predictions can be made through energy simulations and carbon calculations. However, the risks' impacts on the retrofit performance gaps, considering the designed and applied retrofits, can be revealed through data-driven predictions. This study aims to develop a data-driven and risk-based decision support system for enhancing the retrofit performance of buildings. The retrofit performance gaps, in light of risks and uncertainties, covered energy, energy-based carbon emissions, and cost-based results in this study. The dataset covered building parameters, retrofit scales, risks and uncertainties, and retrofit performance-based results before and after retrofits. This dataset, which includes information about 71 retrofitted buildings, was created through semi-structured interviews, web-based surveys, on-site reports, and project information. Data-driven models using DT, RF, SVM, and MLP, as a type of ANN, were applied for retrofit performance predictions on this dataset. RF and MLP models generally resulted in higher accuracy with lower variances in this study. The top features found through feature selection in ML predictions were generally related to risks. A web-based demo using the applied ML models was provided through Gradio, as an example for an output prediction related to retrofit performance for the user, considering the risks.

Subject Keywords

Retrofits, Energy performance, Carbon emissions, Risks, Built environment

URI

https://hdl.handle.net/11511/116036

Collections

Graduate School of Natural and Applied Sciences, Thesis