| 摘要: |
| 街道朝向对于街道界面有显著影响,以朝向讨论为中心的街道界面量化研究尚不充分。本文引入关联规则挖掘技术,量化分析东西向与南北向街道界面在形态表现上的差异。研究发现：“东西向”与“南北向”街道在界面形态上差异明显；界面密度以0.8为界,以上的街道大概率(conf≥75%)为东西向,反之为南北向；另在空隙波动、高度波动等指标上亦有差别；以“东西向”和“南北向”为中心,涌现出两个大的街道界面“形态-朝向”关联规则簇群。研究提出一种营造“城市意向”的新思路：由关联规则提取街道界面形态指标管控范围,凸显“东西向”与“南北向”街道差别,引导人们构建“界面-朝向-定位”心智地图,提升街道空间场所的可意向性。 |
| 关键词: 街道界面 关联规则 朝向 形态 可意向性 |
| DOI： |
| 分类号:TU981 |
| 基金项目:同济大学建筑与城市规划学院自然资源部国土空间文化遗产保护与再生工程技术创新中心开放课题 |
|
| Emergence of "Form-Orientation" Association Rules in Street Interface |
|
LU Mingyang
|
|
Nanjing Tech University
|
| Abstract: |
| Street orientation significantly influences the morphological expression of street interfaces. Due to the limitations of conventional statistical methods, current quantitative research on street interfaces in urban design primarily focuses on defining and calculating indicators based on continuous data, with limited discussion on discrete data such as street orientation. The quantitative study of how street orientation affects its interface form remains insufficient. Research on street orientation needs to break through existing methodological frameworks by introducing Association Rules mining algorithm to analyze "form-orientation" mixed data, exploring the differences in morphological expression between east-west and north-south street interfaces.
In terms of research subjects, the study selected 46 small to medium scale residential streets in the old city of Nanjing as samples. Multi- source data collection was conducted to gather vector data on motorized roadways, pedestrian spaces, and street-front buildings. The study then defined and quantified 15 indicators across five categories—orientation, interface form, composite form, lanes, and functions—completing the systematic analysis.
In terms of research methodology, the study first conducts a detailed analysis of the adaptability of Association Rules in urban design scenarios. By comparing the regression line calculation in scatter plots with the regional division of relationships in Association Rules, it reveals the fundamental mathematical principles of Association Rules in data mining. Subsequently, based on the characteristics of urban design scenarios and previous few experiences, the study establishes the criteria for strong Association Rules: Confidence level (conf) ≥ 0.7 and Support level (supp) ≥ 0.15. Finally, the study employs the Apriori algorithm to complete the mining of strong Association Rules among indicators and presents the primary Association Rule results in a tabular format.
In terms of result analysis, the study first constructed a correlation matrix and orientation divided box plots among the indicators as references for Association Rule analysis. Then, focusing on the most universally applicable Association Rules consisting of 1 LHS and 1 RHS, it visually illustrated the interrelationships between indicators through a honeycomb network formed by connecting the two components of association rules. During this process, the indicators formed two major clusters around the east-west and north-south orientations of streets, revealing the "form-orientation" association patterns in street interfaces. The results of Association Rules aligned with the trends observed in the matrix and box plots, while the Association Rules provided deeper insights into specific intervals within the indicators, demonstrating superior precision.
The study further discussed the practical manifestations, causes, and practical applications of the relationship between "form-orientation" of street interfaces, including empirical examples of typical street view where "form" is associated with "orientation"; The emergence of the phenomenon of "form" being associated with "orientation"; Feasibility exploration of enhancing "imageability" through "orientation".
Research has found that there are significant differences in the interface form between "east-west" and "north-south" streets; The interface density is bounded by 0.8, and streets above that are highly likely to be east-west (conf ≥ 75%), and vice versa are north-south; There are also differences in indicators such as gap fluctuations and height fluctuations; Two large clusters of street interface "form orientation" Association Rules have emerged, centered around the "east-west" and "north-south" directions. A new approach to creating "urban intention" has been proposed: extracting the control range of street interface form indicators from Association Rules, highlighting the differences between "east-west" and "north-south" streets, guiding people to construct a "interface orientation positioning" mental map, and enhancing the meaningfulness of street spatial places.
Although Honeycomb network diagram drawn based on Association Rules have good readability, as a simple knowledge graph, there is still a gap compared to mature research in terms of data sample and category scale, and other indicators. How to mine vertical knowledge graphs in the field of urban design that designers can understand, comprehend, and effectively utilize in complex urban data will be the main challenge for future research. |
| Key words: Street Interface Association Rules Orientation Form Imageability |
