Identifying localized amenities for gentrification using a machine learning-based framework

doi:10.1016/j.apgeog.2022.102748

2022-08-01

PRE

DOI：10.1016/j.apgeog.2022.102748

原文链接

原文求助

摘要

En 中文

The process of gentrification changes the composition and character of urban neighbourhoods in cities worldwide. Amenities such as art galleries, designer boutiques, fine dining, and specialty cafe ' s interact with most gentrification processes and could act as indicators for measuring gentrification. Previous literature has explored the role of amenities in gentrification, and some have found distinctive amenity landscapes in different spatial contexts. However, there is a lack of a more generalized approach for identifying gentrification-related amenities across different regions. This study proposed a machine learning-based framework to identify localized gentrification amenities. Specifically, amenities were represented by Points of Interest (POIs) and matched to the North American Industry Classification System (NAICS), an industry classification system commonly used in amenityrelated studies. Bridging POI categories and the NAICS hierarchy enables a dialog between big data and conventional statistical data. Then, given typical gentrification neighborhoods in an area, featured amenities can be identified via a supervised gradient boosting method. The framework was applied to Shenzhen, a major Chinese city. Results showed that Shenzhen has a distinct amenity landscape in its gentrified neighborhoods; for example, bubble tea beverage shops were recognized as a dominant amenity, as opposed to the cafe ' s in many Western cities, as well as financial institutions, digital electronics, and car-related amenities. The proposed machine learning-based framework not only provides a generalized approach to identifying gentrification-related amenities in different regions, but also enables dynamic and fine-grained tracking of gentrification on the basis of big data.

Keyword:

Localized amenities

Gentrification

Machine learning

POIs