This study investigates the impact of high-speed rail (HSR) on Thailand’s public transportation market and evaluates the effectiveness of machine learning techniques in predicting travel mode choices. A stated preference survey was conducted with 3200 respondents across 16 provinces, simulating travel scenarios involving buses, trains, airplanes, and HSR. The dataset, consisting of 38,400 observations, was analyzed using the CatBoost model and the multinomial logit (MNL) model. CatBoost demonstrated superior predictive performance, achieving an accuracy of 0.853 and an AUC of 0.948, compared to MNL’s accuracy of 0.749 and AUC of 0.879. Shapley additive explanations (SHAP) analysis identified key factors influencing travel behavior, including cost, service frequency, waiting time, travel time, and station access time. The results predict that HSR will capture 88.91% of the intercity travel market, significantly reducing market shares for buses (4.76%), trains (5.11%), and airplanes (1.22%). The findings highlight the transformative role of HSR in reshaping travel patterns and offer policy insights for optimizing pricing, service frequency, and accessibility. Machine learning enhances predictive accuracy and enables a deeper understanding of mode choice behavior, providing a robust analytical framework for transportation planning.
Stored only when both papers are in the corpus — 7 citations exist upstream.
Imported from OpenAlex (CC0) — nothing here is generated.