Urban Functional Area Recognition Based on Unbalanced Clustering

Abstract

Urban functional area recognition refers to refining the main functions of building coverage areas. At present, multisource data analysis is prone to data imbalance, and types with large data volume are more likely to affect data analysis results. Therefore, this study took the main urban area of Taiyuan as the research object and used the Synthetic Minority Oversampling Technology (SMOTE) method to reduce the impact of data imbalance. In this study, the SOMTE method was used to incrementally process the microblog check-in data in the main urban area of Taiyuan, which reduced the phenomenon of data imbalance and further improved the recognition accuracy. The Point of Interest (POI) data were clustered through K-nearest neighbor, and microblog check-in data were semantically analyzed by Linear Discriminant Analysis (LDA). Then, the eigenvalues of the two kinds of data results were obtained by frequency density analysis. Finally, feature fusion was carried out by means of weighted average. The fused data were divided into single and mixed functional areas according to the difference of frequency density, which was rendered and displayed on the ArcGIS platform, so as to realize the visual identification and division of urban functional areas, and the results were compared with Gaode Map. The experimental results showed that this method can effectively identify urban functional areas with a recognition accuracy of 85%, which provided reference value for the planning and research of urban functional areas in the future.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 7 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.