Communication With Detroit: Machine Learning in Open Source Community Housing Design

Abstract

Traditional pre-design investigation includes conceptual studies, site analysis, and programming processes to analyze the site and design for users. Instead, designers and architects should consider users' ideas and their actual usage of space, which are recorded and reflected on the social media platform. To introduce more citizens' voices in the design and learn more about people's expression of Detroit city and its housing, we propose to involve the machine learning analysis in the earlier stage of the housing project using users' reflections from social media to support the conceptual design. This paper introduces a novel design framework that deals with the lacking public programs in Detroit using an online data clustering platform and demonstrates a conceptual open-source community housing design according to related findings. This framework incorporates data collection from the Twitter platform, implementation of clustering for user-oriented programs, and design applications based on the findings. Our research demonstrates an efficient and flexible approach to the open-source community housing project.