SDG scoring at building-level for Hong Kong using Big Data and Machine Learning approach

Existing neighborhood sustainability assessment (NSA) tools use a traditional triple-bottom-line methodology that places an imbalanced and non-spatial focus on sustainability dimensions. This results in a lack of actionable insights for stakeholders, government, urban planners and the general public to make decisions based on their needs and priorities. Presently, spatial inequity of infrastructure access is observed at building-level, neighborhood level and district level. However, the spatial inequity associated with the Sustainable Development Goals (SDGs) and, in particular, their respective indicators are not well understood at the neighborhood scale to inform urban design.

To overcome this problem, this study focuses on developing an algorithm purely based on machine learning and geospatial models to compute and assess the sustainability scorings at the building level. The algorithm utilizes automated data scrapping from online open data portals and other resources, which are cleaned and preprocessed in the next step. The data points include facilities, census and statistics data at building level, services and other resources. The resultant list of individual buildings is mapped, with all 17 sustainable development goals, which breaks down to 232 unique target indicators. We select those indicators that are pertinent to urban systems for the calculation of the scores for each SDG. An aggregated final SDG score can be used to compare buildings in a neighborhood in terms of their overall score that takes into account the performance for each SDG.

The methodology of this paper is focused on the entirety of Hong Kong. However, this methodology can also be applied to other cities and/or countries, given that data points are available as model input. The developed methodology aims to address individual SDG status at the building level, which can also be grained at the neighborhood or district level. This will help to understand how sustainable individual buildings are and provide guidelines to access neighborhood sustainability. The resultant scoring will help to attain overall higher SDG scoring for all buildings by understanding the hotspots for improvement. The individual scoring will help to analyze the current sustainability aspect of buildings and will help in the decision-making of suitable actions to enhance the overall SDG of the area. This will act as a guideline for policy makers and urban planners to understand where and how to improve overall sustainability of a neighborhood.