Research And Assess Of Innovative City Based On Kernel Principal Component Analysis

With the development of society,innovation,as the basic driving force for economic development in countries and regions,has received more and more attention.Innovative cities refer to cities that are mainly driven by innovation factors such as science and technology,knowledge,human resources,culture,and institutions.Building innovative cities is not only a key step in accelerating the country's new urbanization process and building an innovative country,but also exploring new urban development.An important part of the model to promote sustainable urban development.Therefore,the construction of innovative cities reflects the gap in the ability of each city to innovate and the ability to innovate among cities,help cities identify their own positioning,determine the goals for later development,and catch up with them,and determine the city's overall development strategy and macro-control policies very important meaning.This article takes 18 major cities in our country as the research object.Firstly,according to the research literature of innovative cities at home and abroad,the connotation of innovative cities and the existing evaluation index system are combed,and the method of nuclear principal component analysis is selected to conduct comprehensive evaluation of innovative cities;second,follow the system,concise The basic principles of independence,independence,and representativeness,feasibility,and measureability have established an innovative urban evaluation index system that is easily accessible and systematically accessible to the general public.Next,according to different core functions,we evaluated the innovative capabilities of core principal component analysis in 18 major cities in China.We compared the cumulative contribution rates of polynomial kernel functions and Gaussian kernel functions under different parameters,and determined the optimal kernel functions and Based on the optimal nuclear parameters,a new mixed kernel function was reconstructed,and the comprehensive evaluation results of nuclear principal component analysis were compared with the comprehensive evaluation results of principal component analysis.The results show that the mixed kernel function not only combines the advantages of global polynomial kernel function and locality of Gaussian kernel function,but also makes the evaluation result more reasonable.The top five cities in comprehensive evaluation of innovative cities in China are Beijing,Shenzhen,Shanghai and Guangzhou respectively.In Hangzhou,the five cities with the most comprehensive evaluation of innovative cities are Zhengzhou,Dalian,Changsha,Chongqing,and Shenyang.Although the principal component analysis can play a certain dimensionality reduction effect on the multi-index data,the dimensionality reduction effect is obvious.It is not as significant as the analysis of the nuclear principal component,and the evaluation result of the data kernel principal component analysis with a nonlinear relationship is more reasonable than the principal component analysis.Finally,based on the analysis of the results of the comprehensive evaluation of China's innovative cities,the problems existing in China's innovative cities are proposed and four policy recommendations are proposed.