Abstract: Objective The integration of mainline railway, intercity railway, suburban railway and urban rail transit networks（abbreviated as four-network integration） has become a key strategy for optimizing urban transport structure and reducing travel costs through integrating rail transit and road network resources. Therefore, it is necessary to conduct an analysis about the four-network integration travel costs. Method Taking Shanghai as the research object, a travel choice model based on XGBoost is constructed, and the SHAP（Shapley additive explanations） method is applied to interpret the influencing factors in travel decisions. The mechanism of four-network integration on residents’ travel choices is systematically analyzed, and its impact on travel behavior is quantitatively assessed. Through a combination of 1 km×1 km gridded demographic data, multi-source transportation data fusion, and machine learning modeling, the total travel time, off-peak travel time, cost, in-transit time, and physical exertion are selected as key influencing factors. Result ＆ Conclusion An empirical study at Shanghai Hongqiao Hub shows that four-network integration can improve per capita travel efficiency by 20.7%, playing a key role in enhancing the hub’s regional influence and optimizing the travel structure of urban agglomeration. The effectiveness of the XGBoost machine learning framework in assessing the policy impacts of four-network integration is demonstrated, providing theoretical and empirical support for policy formulation and optimization in rail transit four-network integration.