skip to primary navigationskip to content
 

Machine Learning and AI in the Built Environment - Dr Thies Lindenthal

This project will improve the foundatons for applying tried-and-tested machine learning approaches to the built environment. It will reduce the cost of creating and deploying AI systems, facilitating new academic research on e.g. property markets and investments. A future version of the API could be commercialised in areas like mortgage origination, insurance claim processing or property tax (non-UK, though) estimation.

The UK is in an exceptional position when analysing the built environment from an economic perspective: Property transactions for residential real estate are readily available public data (Land Registry, 2017)⁠. However, the level of detail for each transaction is very low. Only basic building attributes are recorded: Property type, freehold status, newly built vs. re-sale. Over the last year, I have utilised Big Data and advances in machine learning to augment these sales data using remotely sensed information, e.g. deriving building size and volume from LIDAR (Lindenthal, 2017a)⁠ or architectural homogeneity from 3D city models (Lindenthal, 2017b)⁠.

A new working paper established a method to extract pictures of individual buildings from Google Streetview (previous research has been constrained to the street or block level). Using deep convolutional neural networks we built a machine learning model for detecting the buildings’ vintage (Georgian/Victorian/…) from these pictures. We were able to classify all of Cambridge’s buildings and to estimate price premia for certain styles (yes, historic buildings command a higher price).

The suggested project will establish the infrastructure necessary to scale this work to the national level and also to increase the level of detail. Additional attributes that could be inferred from remotely sensed data include property size, number of bedrooms, year of construction, maintenance levels, energy efficiency, location amenities, access to transport, replacement values, or even socio-demographic attributes of the residents.

To scale up, three basic infrastructure components are needed: 

  1. Collect expert input efficiently: Models need to be trained on human classified data (training data). This project will provide tool(s) to collect and manage training data from a diverse group of human experts. An intuitive user interface to enable non-technical users to collect pictures of buildings (import, device camera, Google Streetview), to define and manage classifications, to classify pictures efficiently and to evaluate results after automatic classifications.
  2. Store and display: A scalable cloud-based system to store training data, cache derived building data and to visualise results.    
  3. Integrate with other software/apps: An API can be called from external applications to estimate building level attributes based on a dynamic set of models. Two forms of inputs are catered for: An UK address for which pictures will be captured from e.g. Google Streetview or other sources. Alternatively, pictures of buildings can be sent to the API directly. Open source APIs for machine learning models already exist (e.g. Tensorflow Serving) which support individual requests and batch processing.

In sum, this project will improve the foundations for applying tried-and-tested machine learning approaches to the built environment. It will reduce the cost of creating and deploying AI systems, facilitating new academic research on e.g. property markets and investments. A future version of the API could be commercialised in areas like mortgage origination, insurance claim processing or property tax (non-UK, though) estimation. 

References

  • Land Registry, 2017. Price Paid Data [WWW Document]. URL http://landregistry.data.gov.uk/app/ppd
  • Lindenthal, T., 2017a. Estimating Supply Elasticities for Residential Real Estate in the United Kingdom, Comprehens. ed, Reference Module in Earth Systems and Environmental Sciences. Elsevier Inc. doi:https://doi.org/10.1016/B978-0-12-409548-9.09682-2
  • Lindenthal, T., 2017b. Beauty in the Eye of the Home-Owner: Aesthetic Zoning and Residential Property Values. Real Estate Econ. 1–26. doi:10.1111/1540-6229.12204

Researchers:


Land Economy, Department of Economics


University of Alabama
Department of Economics

 

Filed under:

Welcome to the Centre for Digital Built Britain.  

The Centre for Digital Built Britain is a partnership between the Department of Business, Energy & Industrial Strategy and the University of Cambridge to deliver a smart digital economy for infrastructure and construction for the future and transform the UK construction industry’s approach to the way we plan, build, maintain and use our social and economic infrastructure.