Using Machine Learning Techniques to Predict Esthetic Features of Buildings

Aydin, Yusuf Cihat; Mirzaei, Parham A.; Hale, Jonathan

Abstract

Several substantial market barriers obstruct the widespread adoption of sustainable buildings. Esthetic features are amongst the main driving forces behind the marketability of buildings, thus improvement of sustainable buildings in terms of visual esthetics would enhance their marketability and thus their market intake. Nonetheless, esthetic improvement of the buildings is a challenging task because it lacks in scales and methods to measure and evaluate buildings’ facade esthetic. In this regard, this study aims to develop machine learning-based models to predict the esthetic appreciation of buildings related to their façade features. For this purpose, an artificial neural network and decision tree models are developed and validated with the results of a conducted comprehensive survey (n = 807). In addition, the impact of different window features (i.e., position, number, area, width, height, symmetry, and proportion) on housings esthetic and marketability is investigated. Results show a high level of accuracy for both models in the prediction of esthetic appreciation of buildings.

Citation

Aydin, Y. C., Mirzaei, P. A., & Hale, J. (2021). Using Machine Learning Techniques to Predict Esthetic Features of Buildings. Journal of Architectural Engineering, 27(3), https://doi.org/10.1061/%28ASCE%29AE.1943-5568.0000477

• Journal Article Type: Article
• Acceptance Date: Mar 17, 2021
• Online Publication Date: Jun 11, 2021
• Publication Date: Sep 1, 2021
• Deposit Date: Oct 22, 2024
• Journal: Journal of Architectural Engineering
• Print ISSN: 1076-0431
• Electronic ISSN: 1943-5568
• Publisher: American Society of Civil Engineers
• Peer Reviewed: Peer Reviewed
• Volume: 27
• Issue: 3
• DOI: https://doi.org/10.1061/%28ASCE%29AE.1943-5568.0000477
• Keywords: Visual Arts and Performing Arts; Civil and Structural Engineering; Architecture ; Building and Construction
• Public URL: https://nottingham-repository.worktribe.com/output/5671278
• Publisher URL: https://ascelibrary.org/doi/10.1061/%28ASCE%29AE.1943-5568.0000477
• Additional Information: Received: 2020-10-06; Accepted: 2021-03-17; Published: 2021-06-11