Advertisement

This Room Is Too Dark and the Shape Is Too Long: Quantifying Architectural Design to Predict Successful Spaces

  • Carlo BaileyEmail author
  • Nicole Phelan
  • Ann Cosgrove
  • Daniel Davis
Chapter

Abstract

Historically, architects have relied primarily on rules-of-thumb to layout offices. In this paper we consider whether these assumptions can be improved by using machine learning to predict the success of an office layout. We trained a support vector classifier on data from 3276 private offices from 140 buildings. 56 features of the offices were used, including whether it had a window and the office’s squareness. The model was able to predict the lowest performing offices with a precision of 60–70% and a recall between 20 and 40%. This research suggests that many of the assumptions that drive architects will be able to be validated or refuted by applying machine learning to data gathered from people inhabiting the built environment.

Keywords

Machine learning Predictive model Office design Building performance Support vector classifier 

References

  1. Cortes, C., Vapnik, V.: Support-vector networks. Mach Learn 20, 273–297 (1995)Google Scholar
  2. Duffy, F.: Organizational design. J. Archit. Res. 6, 4–9 (1977)Google Scholar
  3. Fisk, W.J., Rosenfield, A.H.: Estimates of improved productivity and health from better indoor environments. Indoor Air 7, 160–172 (1997)Google Scholar
  4. Goldstein, R., Tessier, A., Khan, A.: Space layout in occupant behaviour simulation. In: Proceedings of Building Simulation 2011: 12th Conference of International Building Performance Simulation Association, Sydney (2011)Google Scholar
  5. Harvard Business Review: https://hbr.org/2014/10/workspaces-that-move-people (2014). Last Accessed 15 May 2017
  6. Hastie, T., Tibshirani, R., Friedman, J.: The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Springer Series in Statistics, 2nd edn. https://statweb.stanford.edu/~tibs/ElemStatLearn/printings/ESLII_print10.pdf. As of May 2017
  7. Hillier, B., Leaman, A., Stansall, P., Bedford, M.: Space syntax. Environ. Plan. Des. 3(2), 147–185 (1976)Google Scholar
  8. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., Duchesnay, E.: Scikit-learn: machine learning in python. J. Mach. Learn. Res. 12, 2825–2830 (2012)Google Scholar
  9. Penn, A., Turner, A.: Space syntax based agent simulation. http://discovery.ucl.ac.uk/2027/1/penn.pdf. As of May 2017
  10. Phelan, N., Daniel, D., Carl, A.: Evaluating architectural layouts with neural networks. In: Symposium on Simulation for Architecture and Urban Design (2016)Google Scholar
  11. Tabak, V.: User simulation of space utilisation: system for office building usage simulation. Ph.D. thesis, Eindhoven University of Technology, Netherlands (2009)Google Scholar
  12. The New York Times.: http://www.nytimes.com/2013/03/16/business/at-google-a-place-to-work-and-play.html (2013). Last Accessed 15 May 2017
  13. Van der Voordt, T.J.M.: Productivity and employee satisfaction in flexible workplaces. J. Corp. Real Estate. 6(2) (2003)Google Scholar

Copyright information

© Springer Nature Singapore Pte Ltd.  2018

Authors and Affiliations

  • Carlo Bailey
    • 1
    Email author
  • Nicole Phelan
    • 1
  • Ann Cosgrove
    • 1
  • Daniel Davis
    • 1
  1. 1.New YorkUSA

Personalised recommendations