[Objective] Improving public transport services is essential for promoting sustainable travel and reducing dependence on private cars. Perceived accessibility reflects travelers’ subjective assessment of how easily they can accomplish their travel purposes using the public transport system and is closely associated with both travel mode choice and perceived service quality. As such, it serves as a critical link between perceived service quality and modal share. However, previous studies have rarely identified the service attributes associated with perceived accessibility while simultaneously examining how modal share may change under different hypothetical levels of perceived accessibility. This study developed an integrated framework to identify service-related pathways influencing perceived accessibility and to simulate modal share responses. [Methods] This study used data from the 2023 Hangzhou Household Travel Survey and a questionnaire evaluating public transport services and user satisfaction. After data screening, 5 135 respondents and 12 530 trip records were retained. A partial least squares structural equation model (PLS-SEM) was first developed to identify service-related pathways influencing perceived accessibility. The model incorporated psychological latent variables representing key dimensions of public transport service, including comfort, reliability, information availability, and safety. The estimated perceived accessibility scores were then incorporated into a random forest (RF) model, together with sociodemographic and trip characteristics, to predict travel mode choice. Shapley additive explanations (SHAP) analysis evaluated the contribution of perceived accessibility to the model predictions. Scenarios were subsequently constructed by progressively increasing perceived accessibility to simulate changes in modal share and examine potential substitution between public transport and private cars. Model training and scenario simulations were repeated using 20 random seeds to assess the robustness of the results. Finally, heterogeneity analyses by trip distance and driver’s license status were conducted to investigate differences across traveler groups. [Results] The PLS-SEM measurement model demonstrated strong reliability and validity. Comfort, reliability, information availability, and safety were all significantly and positively associated with perceived accessibility. Reliability exhibited both the strongest direct effect (standardized path coefficient 0.279) and the largest total effect (0.315), suggesting that service improvements should prioritize operational stability, transfer coordination, and on-time performance, particularly during peak periods. The RF model achieved macro-average and weighted-average F1 scores of 0.792 and 0.812, respectively, indicating satisfactory predictive performance. SHAP analysis ranked perceived accessibility as the third most influential predictor of travel mode choice. As perceived accessibility increased, the predicted share of public transport increased steadily, whereas the share of private car travel declined substantially. However, the responsiveness of modal share diminished beyond a certain threshold. Heterogeneity analyses further showed that the increase in public transport use and the decrease in private car use were more pronounced for medium- and long-distance trips than for short-distance trips. Travelers with a driver’s license also exhibited stronger responses than those without one. These findings suggested that improving perceived accessibility alone might not be sufficient to consistently increase public transport use. Public transport policies should therefore prioritize service improvements for medium- and long-distance travel while coordinating these efforts with complementary measures, such as parking management. [Conclusions] By integrating PLS-SEM and RF models, this study identifies service-related pathways influencing perceived accessibility and simulates modal share responses under different improvement scenarios. The findings clarify the bridging role of perceived accessibility between public transport service quality and travel behavior. They provide an empirical basis for developing targeted strategies to improve public transport services and offer practical guidance for encouraging shifts from private cars to public transport, thereby promoting sustainable, low-carbon urban mobility.
Key words
travel mode choice /
perceived accessibility /
public transport /
partial least squares structural equation model /
machine learning models
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
References
[1] TEJASWI S P, BALIJEPALLI C, PRASAD C S R K. Revealing commute choice factors: SEM analysis of public transport and active modes in Hyderabad, India[J]. Journal of Urban Planning and Development, 2024, 150(3): 04024026.
[2] QIN X R, KE J T, WANG X L, et al. Demand management for smart transportation: A review[J]. Multimodal Transportation, 2022, 1(4): 100038.
[3] LÄTTMAN K, OLSSON L E, FRIMAN M. Development and test of the perceived accessibility scale (PAC) in public transport[J]. Journal of Transport Geography, 2016, 54: 257-263.
[4] FRIMAN M, LÄTTMAN K, OLSSON L E. Public transport quality, safety, and perceived accessibility[J]. Sustainability, 2020, 12(9): 3563.
[5] CHAU H W, GAISIE E, JAMEI E, et al. Perceived accessibility: Impact of social factors and travel modes in Melbourne’s west[J]. Applied Sciences, 2024, 14(15): 6399.
[6] CHAU H W, CHAN M, JAMEI E, et al. The impacts of perceived safety and service quality on perceived accessibility by public transport in Melbourne[J]. Land, 2024, 13(11): 1928.
[7] GUIRAO B, GARCÍA-PASTOR A, LÓPEZ-LAMBAS M E. The importance of service quality attributes in public transportation: Narrowing the gap between scientific research and practitioners’ needs[J]. Transport Policy, 2016, 49: 68-77.
[8] WATTHANAKLANG D, JOMNONKWAO S, CHAMPAHOM T, et al. Exploring accessibility and service quality perceptions on local public transportation in Thailand[J]. Case Studies on Transport Policy, 2024, 15: 101144.
[9] PINEDA-JARAMILLO J, ARBELÁEZ-ARENAS Ó. Assessing the performance of gradient-boosting models for predicting the travel mode choice using household survey data[J]. Journal of Urban Planning and Development, 2022, 148(2): 04022007.
[10] HE M W, LI J B, SHI Z B, et al. Exploring the nonlinear and threshold effects of travel distance on the travel mode choice across different groups: An empirical study of Guiyang, China[J]. International Journal of Environmental Research and Public Health, 2022, 19(23): 16045.
[11] SCHEEPERS C E, WENDEL-VOS G C W, VAN KEMPEN E E M M, et al. Perceived accessibility is an important factor in transport choice: Results from the AVENUE project[J]. Journal of Transport & Health, 2016, 3(1): 96-106.
[12] SHENG L M, ZHANG L B. Understanding the determinants for predicting citizens’ travel mode change from private cars to public transport in China[J]. Frontiers in Psychology, 2022, 13: 1007949.
[13] NEGM H, EL-GENEIDY A. Subjectivity matters: Investigating the relationship between perceived accessibility and travel behaviour[J]. Transportation Research Part A: Policy and Practice, 2025, 193: 104399.
[14] KIM J, LEE B. Campus commute mode choice in a college town: An application of the integrated choice and latent variable (ICLV) model[J]. Travel Behaviour and Society, 2023, 30: 249-261.
[15] 高捷, 李丹, 张士行. 心理潜变量与微观环境融合视角下的短途绿色出行选择研究[J/OL]. 交通运输系统工程与信息. (2026-06-01) [2026-07-03]. https://link.cnki.net/urlid/11.4520.U.20260601.1142.008. GAO J, LI D, ZHANG S X. Short-distance green travel choice considering psychological latent variables and micro- environmental context[J/OL]. Journal of Transportation Systems Engineering and Information Technology. (2026-06-01) [2026- 07-03]. https://link.cnki.net/urlid/11.4520.U.20260601.1142.008. (in Chinese)
[16] LI X W, WANG Y T, WU Y, et al. Modeling intercity travel mode choice with data balance changes: A comparative analysis of Bayesian logit model and artificial neural networks[J]. Journal of Advanced Transportation, 2021, 2021: 9219176.
[17] AJZEN I. The theory of planned behavior[J]. Organizational Behavior and Human Decision Processes, 1991, 50(2): 179-211.
[18] PARASURAMAN A, ZEITHAML V A, BERRY L L. SERVQUAL: A multiple-item scale for measuring consumer perceptions of service quality[J]. Journal of Retailing, 1988, 64(1): 12-40.
[19] ZHANG C Q, LIU Y, LU W T, et al. Evaluating passenger satisfaction index based on PLS-SEM model: Evidence from Chinese public transport service[J]. Transportation Research Part A: Policy and Practice, 2019, 120: 149-164.
[20] LÄTTMAN K, FRIMAN M, OLSSON L E. Perceived accessibility of public transport as a potential indicator of social inclusion[J]. Social Inclusion, 2016, 4(3): 36-45.
[21] HAIR J F JR, HULT G T M, RINGLE C M, et al. Partial least squares structural equation modeling (PLS-SEM) using R: A workbook[M]. Cham: Springer, 2021.
[22] HAIR J F, RISHER J J, SARSTEDT M, et al. When to use and how to report the results of PLS-SEM[J]. European Business Review, 2019, 31(1): 2-24.
[23] 沈金星, 李麒年, 于淼, 等. 出行链视角下地铁乘客满意度影响因素的组态效应分析[J]. 东南大学学报(自然科学版), 2025, 55(4): 1165-1172. SHEN J X, LI Q N, YU M, et al. Group effect analysis of factors influencing subway passenger satisfaction from a travel chain perspective[J]. Journal of Southeast University (Natural Science Edition), 2025, 55(4): 1165-1172. (in Chinese)
[24] VAN DER VLUGT A L, CURL A, SCHEINER J. The influence of travel attitudes on perceived walking accessibility and walking behaviour[J]. Travel Behaviour and Society, 2022, 27: 47-56.
[25] YUAN Y, SHAO C F, CAO Z C, et al. The effect of travel-chain complexity on public transport travel intention: A mixed-selection model[J]. International Journal of Environmental Research and Public Health, 2023, 20(5): 4547.
[26] LAW L, FONG N. Applying partial least squares structural equation modeling (PLS-SEM) in an investigation of undergraduate students’ learning transfer of academic English[J]. Journal of English for Academic Purposes, 2020, 46: 100884.
[27] JAFARI M, PERSAUD B. Assessing the moderation effects of socioeconomic factors in predicting severe crash outcomes based on road and environment characteristics: A modeling comparison[J]. Transportation Research Record: Journal of the Transportation Research Board, 2026: 03611981261423464.
[28] TAMIM KASHIFI M, JAMAL A, SAMIM KASHEFI M, et al. Predicting the travel mode choice with interpretable machine learning techniques: A comparative study[J]. Travel Behaviour and Society, 2022, 29: 279-296.
[29] HAGENAUER J, HELBICH M. A comparative study of machine learning classifiers for modeling travel mode choice[J]. Expert Systems with Applications, 2017, 78: 273-282.
[30] LUNDBERG S M, LEE S I. A unified approach to interpreting model predictions[C]// Proceedings of the 31st International Conference on Neural Information Processing Systems. Long Beach, USA: Curran Associates Inc., 2017: 4765-4774.
[31] LUNDBERG S M, ERION G, CHEN H, et al. From local explanations to global understanding with explainable AI for trees[J]. Nature Machine Intelligence, 2020, 2(1): 56-67.
[32] 李文权, 邓安鑫, 郑炎, 等. 基于机器学习的中型城市居民出行方式选择行为研究[J]. 交通运输系统工程与信息, 2024, 24(2): 13-23. LI W Q, DENG A X, ZHENG Y, et al. Analysis of residents’ travel mode choice in medium-sized city based on machine learning[J]. Journal of Transportation Systems Engineering and Information Technology, 2024, 24(2): 13-23. (in Chinese)
[33] XU Y, WANG Y K, LIU H L, et al. Examining the nonlinear relationship between built environment and residents’ leisure travel distance: A case study of Guangzhou, China[J]. Land, 2025, 14(12): 2392.
[34] ZAPATA-DIOMEDI B, BOTH A, ABBAS A, et al. Shifting car travel to active modes to improve population health and achieve transport goals: A simulation study[J]. Journal of Transport & Health, 2023, 31: 101628.