Integrating landscape metrics and socioeconomic features for urban functional region classification. (November 2018)

• Record Type: Journal Article
• Authors: Xing, Hanfa; Meng, Yuan
• Is Part Of: Computers, environment and urban systems. Volume 72(2018)
• Page Start: 134
• Page End: 145
• Publication Date: 2018-11
• Subjects: Urban functional region -- Classification -- Landscape metrics -- Crowdsourced data -- Socioeconomic features; City planning -- Data processing -- Periodicals; Regional planning -- Data processing -- Periodicals; 303.4834
• DOI: 10.1016/j.compenvurbsys.2018.06.005
• Languages: English
• ISSNs: 0198-9715

Abstract

Abstract: Urban functional region classification plays an important role in urban planning, resource management and sustainable development. While previous studies have solely focused on delineating regions by urban morphology or deriving functional types from human activities, the sufficient depiction and integration of these characteristics are seldom discussed. This paper inherently considers both urban morphology and human activities to classify urban functional regions. Building-based and region-based landscape metrics derived from building-level blocks are delineated to measure urban morphology, while socioeconomic features are extracted from crowdsourced data related to human activities using a topic model and semantic scaling method. These characteristics are then fused by applying random forest to measure different functions. This approach is tested in Futian district, Shenzhen, Guangdong. The result shows an overall classification accuracy of 0.818, approximately 0.15 higher than that utilising only landscape metrics or only socioeconomic features. This result indicates the effectiveness of the delineated characteristics to depict urban landscapes and socioeconomic information and the reliability of integrating these features for urban functional region classification. Highlights: An urban functional region classification approach is presented involving landscape metrics and socioeconomic features. Building-based and region-based landscape metrics are delineated from building-level blocks. Socioeconomic features are extracted and semantically scaled from crowdsourced data. Comparative experiments with landscape metrics or socioeconomic features prove the effectiveness of theproposed approach.