Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model

Within globalization, the significance of urban innovation cooperation has become increasingly evident. However, urban innovation cooperation faces challenges due to various factors—social, economic, and spatial—making it difficult for traditional methods to uncover the intricate nonlinear relations...

Full description

Saved in:
Bibliographic Details
Main Authors: Rui Wang, Xingping Wang, Zhonghu Zhang, Siqi Zhang, Kailun Li
Format: Article
Language:English
Published: MDPI AG 2025-03-01
Series:Systems
Subjects:
Online Access:https://www.mdpi.com/2079-8954/13/3/187
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850280190816550912
author Rui Wang
Xingping Wang
Zhonghu Zhang
Siqi Zhang
Kailun Li
author_facet Rui Wang
Xingping Wang
Zhonghu Zhang
Siqi Zhang
Kailun Li
author_sort Rui Wang
collection DOAJ
description Within globalization, the significance of urban innovation cooperation has become increasingly evident. However, urban innovation cooperation faces challenges due to various factors—social, economic, and spatial—making it difficult for traditional methods to uncover the intricate nonlinear relationships among them. Consequently, this research concentrates on cities within the Yangtze River Delta region, employing an explainable machine learning model that integrates eXtreme Gradient Boosting (XGBoost), SHapley Additive exPlanations (SHAP), and Partial Dependence Plots (PDPs) to investigate the nonlinear and interactive effects of multidimensional factors impacting urban innovation cooperation. The findings indicate that XGBoost outperforms LR, SVR, RF, and GBDT in terms of accuracy and effectiveness. Key results are summarized as follows: (1) Urban innovation cooperation exhibits different phased characteristics. (2) There exist nonlinear and interactive effects between urban innovation cooperation and multidimensional factors, among them, the Scientific and Technological dimension contributes the most (30.59%) and has the most significant positive promoting effect in the later stage after surpassing a certain threshold. In the Social and Economic dimension (23.61%), the number of Internet Users (IU) contributes the most individually. The Physical Space dimension (20.46%) generally exhibits mutation points during the early stages of urban development, with overall relationships predominantly characterized by nonlinear positive trends. (3) Through the application of PDP, it is further determined that IU has a positive synergistic effect with per capita Foreign Direct Investment (FDI), public library collections per capita (LC), and city night light data (NPP), while exhibiting a negative antagonistic effect with Average Annual Wage of Staff (AAS) and number of Enterprises above Designated Size in Industry (EDS). (4) For cities at different developmental stages, tailored development proposals should be formulated based on single-factor contribution and multifactor interaction effects. These insights enhance our understanding of urban innovation cooperation and elucidate the nonlinear and interactive effects of multidimensional influencing factors.
format Article
id doaj-art-357f7edde15b4cf7a7ae24db87ac0839
institution OA Journals
issn 2079-8954
language English
publishDate 2025-03-01
publisher MDPI AG
record_format Article
series Systems
spelling doaj-art-357f7edde15b4cf7a7ae24db87ac08392025-08-20T01:48:50ZengMDPI AGSystems2079-89542025-03-0113318710.3390/systems13030187Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning ModelRui Wang0Xingping Wang1Zhonghu Zhang2Siqi Zhang3Kailun Li4School of Architecture, Southeast University, Nanjing 210096, ChinaSchool of Architecture, Southeast University, Nanjing 210096, ChinaSchool of Architecture, Southeast University, Nanjing 210096, ChinaGraduate School of Architecture, Planning and Preservation, Columbia University, New York, NY 10027, USASchool of Architecture, Southeast University, Nanjing 210096, ChinaWithin globalization, the significance of urban innovation cooperation has become increasingly evident. However, urban innovation cooperation faces challenges due to various factors—social, economic, and spatial—making it difficult for traditional methods to uncover the intricate nonlinear relationships among them. Consequently, this research concentrates on cities within the Yangtze River Delta region, employing an explainable machine learning model that integrates eXtreme Gradient Boosting (XGBoost), SHapley Additive exPlanations (SHAP), and Partial Dependence Plots (PDPs) to investigate the nonlinear and interactive effects of multidimensional factors impacting urban innovation cooperation. The findings indicate that XGBoost outperforms LR, SVR, RF, and GBDT in terms of accuracy and effectiveness. Key results are summarized as follows: (1) Urban innovation cooperation exhibits different phased characteristics. (2) There exist nonlinear and interactive effects between urban innovation cooperation and multidimensional factors, among them, the Scientific and Technological dimension contributes the most (30.59%) and has the most significant positive promoting effect in the later stage after surpassing a certain threshold. In the Social and Economic dimension (23.61%), the number of Internet Users (IU) contributes the most individually. The Physical Space dimension (20.46%) generally exhibits mutation points during the early stages of urban development, with overall relationships predominantly characterized by nonlinear positive trends. (3) Through the application of PDP, it is further determined that IU has a positive synergistic effect with per capita Foreign Direct Investment (FDI), public library collections per capita (LC), and city night light data (NPP), while exhibiting a negative antagonistic effect with Average Annual Wage of Staff (AAS) and number of Enterprises above Designated Size in Industry (EDS). (4) For cities at different developmental stages, tailored development proposals should be formulated based on single-factor contribution and multifactor interaction effects. These insights enhance our understanding of urban innovation cooperation and elucidate the nonlinear and interactive effects of multidimensional influencing factors.https://www.mdpi.com/2079-8954/13/3/187urban innovation cooperationnonlinear and interactive effectsmultidimensional influencing factorseXtreme Gradient BoostingSHapley Additive exPlanations
spellingShingle Rui Wang
Xingping Wang
Zhonghu Zhang
Siqi Zhang
Kailun Li
Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model
Systems
urban innovation cooperation
nonlinear and interactive effects
multidimensional influencing factors
eXtreme Gradient Boosting
SHapley Additive exPlanations
title Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model
title_full Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model
title_fullStr Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model
title_full_unstemmed Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model
title_short Research on the Nonlinear and Interactive Effects of Multidimensional Influencing Factors on Urban Innovation Cooperation: A Method Based on an Explainable Machine Learning Model
title_sort research on the nonlinear and interactive effects of multidimensional influencing factors on urban innovation cooperation a method based on an explainable machine learning model
topic urban innovation cooperation
nonlinear and interactive effects
multidimensional influencing factors
eXtreme Gradient Boosting
SHapley Additive exPlanations
url https://www.mdpi.com/2079-8954/13/3/187
work_keys_str_mv AT ruiwang researchonthenonlinearandinteractiveeffectsofmultidimensionalinfluencingfactorsonurbaninnovationcooperationamethodbasedonanexplainablemachinelearningmodel
AT xingpingwang researchonthenonlinearandinteractiveeffectsofmultidimensionalinfluencingfactorsonurbaninnovationcooperationamethodbasedonanexplainablemachinelearningmodel
AT zhonghuzhang researchonthenonlinearandinteractiveeffectsofmultidimensionalinfluencingfactorsonurbaninnovationcooperationamethodbasedonanexplainablemachinelearningmodel
AT siqizhang researchonthenonlinearandinteractiveeffectsofmultidimensionalinfluencingfactorsonurbaninnovationcooperationamethodbasedonanexplainablemachinelearningmodel
AT kailunli researchonthenonlinearandinteractiveeffectsofmultidimensionalinfluencingfactorsonurbaninnovationcooperationamethodbasedonanexplainablemachinelearningmodel