Evaluating multimodal transportation’s impact on city attractiveness: A machine learning approach

• A machine learning approach is applied to compare multimodal transportation infrastructure. • HSR is more important than highways and aviation in predicting city attractiveness during 2008–2018 in China. • Threshold effects analysis assists in determining the optimal scale for deploying transportation infrastructure. Intercity transportation infrastructure plays a vital role in promoting urban development by enhancing accessibility. Efficient resource allocation for developing various transportation modes is essential for policymakers. The advent of high-speed rail (HSR) has sparked increased interest in comparing multimodal transportation infrastructures, such as railways, highways, and aviation. Previous studies have examined these systems from multiple perspectives, including cost, operation, modal choice, network structure, and socioeconomic impact. However, their influence on city attractiveness remains unclear. This study aims to address this gap by utilizing machine learning models, including Principal Component Analysis (PCA) and Gradient Boosting Decision Tree (GBDT), to compare the impact of railways, highways, and aviation on city attractiveness. The analysis employs data from 286 prefecture-level cities in China from 2002 to 2018. The results indicate that HSR has a relatively higher importance in predicting city attractiveness compared to highways and aviation, particularly during the rapid development period of HSR from 2008 to 2018 in China. The analysis also reveals the threshold effects of transportation infrastructure on city attractiveness. This study offers valuable insights for policymakers to improve city attractiveness. The findings help prioritize different transportation infrastructures and determine the optimal scale of infrastructure deployment.