A Thesis On The Characteristics And Influence Mechanism Of The Vitality Pattern Of Urban Centers Based On Multi-Source Spatio-Temporal Data

As China’s urban development enters a phase of connotative upgrading,building vibrant cities has become a common goal for cities,and the popularity of big data technology has opened up more possibilities for in-depth study of urban issues and operational mechanisms.Therefore,a number of cities,including Shanghai,have proposed central activity zones and other related planning and policy measures to promote the transformation of urban centers and further enhance urban vitality.However,in the face of the complex state of urban vitality,traditional urban research methods have become difficult to describe accurately.A common challenge for planning scholars is how to efficiently and fully exploit the phenomenon of urban vitality and the mechanisms behind it,and to apply the findings to planning practice.This study attempts to introduce non-negative matrix decomposition,gradient boosting regression,local dependency diagrams and other relevant machine learning methods and ideas based on multi-source spatio-temporal data,to obtain the static characteristics of Shanghai’s built environment,the dynamic characteristics of Shanghai’s vitality pattern,and the interaction mechanism and marginal effects between them.The main findings of the study include: the resources of Shanghai’s built environment are unevenly distributed and the siphoning effect of the inner ring area is obvious;the vitality pattern of Shanghai has six dynamic features on weekdays and five on rest days,and can be divided into seven typical vitality patterns on both weekdays and rest days,and the obtained dynamic features and vitality patterns are verified to be in line with crowd activity patterns;the improvement of spatial patterns and moderate increase of functional mix have positive benefits on the vitality of most districts in Shanghai;there are obvious commonalities and differences in the composition of vitality patterns in Shanghai commercial centers at different development stages and levels,and the findings can be used for further planning decisions.Compared with traditional vitality research methods,the process of vitality analysis proposed in this paper is faster and more efficient,and has a certain generalizability.With the input of multi-source spatio-temporal data generally available to urban researchers,the highprecision machine learning model can be used to delineate the spatio-temporal characteristics of urban vibrancy and quickly screen and identify different vibrancy areas in the city,becoming the target for further planning research adjustment.