Scholarlink

<p>Understanding competition in urban retail networks is essential for effective commercial planning. This study proposes a Voronoi-based community detection method designed for convenience store networks in Johor Bahru, Malaysia. The method constructs a weighted undirected graph using Delaunay triangulation, where the edge weights integrate geographic distance, brand similarity, socioeconomic differences, and local clustering structure. Based on this, communities are identified through a shortest-path Voronoi partitioning framework guided by local density. Our approach supports multiscale analysis by adjusting the radius parameter, revealing retail community structures at micro, meso, and macro levels. The results show that 99 Speedmart consistently dominates on all scales, while other brands such as 7-Eleven, FamilyMart, and MyNews display varied and more selective market strategies. A detailed boundary analysis also identifies high-potential investment zones located at the edges of different communities, where competition overlaps and consumer flows are dense. Compared to baseline algorithms such as Leiden, Walktrap and FastGreedy, the proposed method achieves comparable modularity while providing higher differentiation of commercial zones, more realistic variation in community sizes, and competitive runtime performance, thereby capturing commercial patterns that align more closely with observed retail structures. In general, this method offers both analytical precision and practical value, supporting spatial decision making in retail site planning and market expansion.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Graph-based community detection in convenience store networks using Voronoi partitioning and multi-weights

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月29日

查看全文
  • Yang Jiao,
  • Suhaibah Azri,
  • Uznir Ujang

摘要

Understanding competition in urban retail networks is essential for effective commercial planning. This study proposes a Voronoi-based community detection method designed for convenience store networks in Johor Bahru, Malaysia. The method constructs a weighted undirected graph using Delaunay triangulation, where the edge weights integrate geographic distance, brand similarity, socioeconomic differences, and local clustering structure. Based on this, communities are identified through a shortest-path Voronoi partitioning framework guided by local density. Our approach supports multiscale analysis by adjusting the radius parameter, revealing retail community structures at micro, meso, and macro levels. The results show that 99 Speedmart consistently dominates on all scales, while other brands such as 7-Eleven, FamilyMart, and MyNews display varied and more selective market strategies. A detailed boundary analysis also identifies high-potential investment zones located at the edges of different communities, where competition overlaps and consumer flows are dense. Compared to baseline algorithms such as Leiden, Walktrap and FastGreedy, the proposed method achieves comparable modularity while providing higher differentiation of commercial zones, more realistic variation in community sizes, and competitive runtime performance, thereby capturing commercial patterns that align more closely with observed retail structures. In general, this method offers both analytical precision and practical value, supporting spatial decision making in retail site planning and market expansion.