Research On The Influencing Factors And Relationship Prediction Of Urban Collaborative Innovation Based On Patent Data

China’s economy is gradually shifting towards a stage of innovation driven and high-quality development,with innovation becoming the main factor driving regional economic growth.Cities are the main spatial carriers of innovation,and collaborative innovation can achieve the sharing of innovation resources and complementary advantages of innovation functions between cities.In the context of "flow space",the innovation relationship between cities presents a trend of complexity and networking.This thesis constructs the urban collaborative innovation network based on patent data,and uses social network analysis methods to calculate network density,degree centrality,and analyze the evolution of the structural characteristics and node characteristics of the urban collaborative innovation network.Using the GBDT method studying the influencing factors of urban cooperative innovation.Based on the results of GBDT,this thesis analyzes the relative importance of each influencing factor and draws relationship diagrams between the influencing factors and urban cooperative innovation to explore the interactive effects between the influencing factors.Based on the research results of influencing factors,the Node2 Vec method is applied to learn the network structure characteristics of city nodes in the urban cooperative innovation network and the characteristics of one pot coded city information,and the prediction of urban cooperative innovation relations and cooperation fields is realized through multiple logistic regression.Research shows:(1)The intensity of urban cooperation and innovation continues to increase,and the breadth of cooperation and innovation continues to expand.The urban cooperation and innovation network shows a tight and balanced development.Especially in cities located in the central and eastern regions,the development is relatively rapid.(2)The GBDT nonlinear model can better explain the influencing factors of urban cooperative innovation.Among the three variables of transport infrastructure,innovation environment and city size,high-speed rail shift,science and technology expenditure and talent reserve are the three factors that have the greatest impact on urban cooperative innovation.(3)The impact of high-speed rail on urban collaborative innovation is non-linear and has a threshold effect.High speed rail can effectively alleviate the hindrance of distance to urban collaborative innovation.High speed rail has a synergistic promoting effect with technology expenditures and talent reserves.(4)The Node2 Vec method is used to learn the network structure characteristics of urban nodes in the urban collaborative innovation network.At the same time,the model that integrates one hot encoded urban information features can accurately predict the relationship and cooperation field of urban collaborative innovation,with an accuracy of 0.849 and an F1-Score of 0.914.The prediction accuracy and reliability are high.In order to promote the improvement of urban cooperation and innovation level,and further promote the construction of national innovation driven strategies and regional integration development,based on the conclusions drawn above,this thesis proposes the following policy recommendations: cities should explore new development paths for optimizing cooperation and innovation relationships and areas based on their own conditions and local conditions.