Identifying Inefficient Urban Residential Land within Shenzhen City: An Approach Using Gaussian Mixture Model and Multi-Source Big Data

Abstract. As land resources become increasingly scarce, urban spatial development patterns in Chinese cities are now shifting from incremental expansion to inventory optimization. Accurate identification of inefficient urban residential land is the key for government to make regeneration policy and improve the living environment of more urban residents. With the rapid urbanization process and uneven resource allocation, China currently faces a declining trend in residential land use efficiency, significantly impacting urban residents' quality of life and satisfaction. Previous research mainly analyzed the land use efficiency of individual residential areas, neglecting to discuss the impact of surrounding environment and the utilization of urban dynamic big data. To address these issues, this paper employs the Gaussian Mixture Model (GMM) clustering method and integrates multi-source geographical big data to quantitatively characterize land use efficiency. Additionally, Spearman's coefficient analysis and Principal Component Analysis methods are applied for data dimensionality reduction. This methodology was initially applied in Bao'an District of Shenzhen and then expanded to cover the entire city. The research results validate the effectiveness and robustness of the approach. The study found that the kappa coefficients for inefficient residential communities and inefficient urban village residences are 0.637 and 0.721, respectively. Spatial analysis reveals that inefficient residential communities are dispersed, while inefficient urban village residences are concentrated in specific areas. This outcome provides important guidance for future government strategies on renewing inefficient urban spaces. At the methodological level, the Gaussian Mixture Model (GMM) clustering method, with its objectivity, multi-dimensionality, and precision, offers a new perspective and approach for studying inefficient urban residential land issues.