建成环境对街道活力的非线性效应——基于XGBoost模型的多源大数据分析

深入认知建成环境对街道活力的影响效应，是街道环境优化与城市更新的基础。但多数研究忽略了二者之间的非线性关系，难以有效指导设计实践。基于多源大数据测度街道活力与建成环境，使用机器学习算法来分析其非线性效应，并针对不同类型街道进行探索。结果表明：1)提升开发强度是促进街道活力的最有效措施；2)建成环境要素对街道活力的影响表现出非线性特征，将其控制在合理范围内，街道活力才会有效提升；3)街道环境要素的综合设置应考虑其交互效应，一个要素的影响会随着另一个要素的变化被放大或缩小；4)老城区、商务片区、工业与区域交通设施周边区域及景观性街道的活力形成机制存在显著差异。相关规律可为街道的精细化设计提供人本尺度的理论参考。

Nonlinear Effect of Built Environment on StreetVitality: A Multi-source Big Data Analysis Based onXGBoost Model

Understanding the influence of built environmenton street vitality is the basis of street environment optimizationand urban renewal. However, most studies ignore the nonlinearrelationship between them, which makes it difficult to effectivelyguide the design practice. Based on multi-source big data tomeasure street vitality and built environment, machine learningalgorithm is used to analyze the nonlinear effect, and differenttypes of streets are explored. The results show that: 1) Enhancingdevelopment intensity is the most effective measures to promotestreet vitality; 2) The influence of environmental factors on thestreet vitality shows nonlinear characteristics. Only by controllingthem within reasonable ranges can the street vitality be effectivelyimproved; 3) The comprehensive setting of street environmentelements should consider their interaction effects, and the influenceof one element will be enlarged or reduced with the change ofanother element; 4) There are significant differences in the vitalityformation mechanisms among the streets in the old urban area,the business district, the surrounding areas of industrial lands andregional transportation facilities and the landscape streets. Relevantlaws can provide a theoretical reference of people-oriented scale forthe refined design of streets.