Development of an artificial intelligence model for adaptive reuse decisions of abandoned industrial buildings

Industrial buildings, due to technological advancements and transformations in production models, tend to lose their original functions more rapidly compared to other building types. However, these structures often retain their structural systems and physical integrity, making them suitable candidates for adaptive reuse through appropriate intervention strategies. In this thesis, the adaptive reuse of industrial buildings that have lost their original function and become "abandoned structures" is addressed through a proposed artificial intelligence (AI)-supported model. The developed AI-based model aims to determine the most appropriate new function with minimal intervention by evaluating the physical parameters of these buildings. The model is built upon the Random Forest classification algorithm, processing through measurable physical and environmental parameters. The usability of this model was tested on five industrial buildings with different layouts and locations in the Western Mediterranean Region of Türkiye, including Antalya Battery Factory, Burdur ÇİNEF Rose Oil Factory and its warehouse, Isparta Leather Factory, and Hüseyin Baş Watermill. Additionally, a comparative analysis was conducted between the model's function proposals and the preferences gathered from a survey of 38 architects.This study aims to fill a gap in the literature by integrating AI applications into architectural decision-support processes and to propose an innovative approach for the adaptive reuse of industrial buildings. As an outcome of the research, it is envisaged that the proposed model will be further developed into a mobile application, aligned with advancements in AI technologies, serving as a decision-support tool for public authorities, and design professionals engaged in the adaptive reuse of built heritage.