Info Research on the discrimination of detour behavior and influencing factors of shared bicycles

Abstract

To investigate the distribution characteristics and influencing factors of bicycle detour behavior, this study accurately identified detour behavior using global positioning system (GPS) track data from shared bicycles. Factors such as travel time, road conditions, public transportation facilities, and land use types were considered in constructing a detour behavior influence model based on the CatBoost machine learning algorithm. The interpretability of the machine learning framework was enhanced via Shapley additive explanations (SHAP), enabling an analysis of the impact of each factor on detour behavior. The results indicated that the CatBoost model effectively recognized detour behavior with high accuracy. The frequency of detour behavior increased with higher road levels, greater distances to crossing facilities, wider bike lanes, and an increased number of bus stops, subway stations, and leisure and entertainment facilities, while it decreased with a higher number of office commuting facilities. In addition, detour behavior was more prevalent on weekends, during off-peak hours, and under conditions involving physical central lane separation and physical bike lane separation. These findings offer a novel approach for identifying bicycle riding behaviors and analyzing their influencing factors, providing effective technical support for non-motorized traffic management and infrastructure optimization.

Keywords

shared bicycles / detour behavior / trajectory data / machine learning / influencing factor

References

[1] SHU S N, BIAN Y, RONG J, et al. Process of phased shift from car to bicycle travel based on the transtheoretical model[J]. Journal of Southeast University (Natural Science Edition), 2018, 48(2): 373-380. (in Chinese)

[2] MA D D. Beijing’s shared bicycle rides surpass 1 billion in 2023[N]. Capital Construction News, 2023. (in Chinese)

[3] LIANG X, XIAO Z S, YUAN Y, et al. Study on the conditions and key parameters for the setting of two-way bicycle lanes[J]. Journal of Transportation Systems Engineering and Information Technology, 2016, 16(5): 227-232. (in Chinese)

[4] LIU L, FAN J P. Study on bicycle traffic planning based on ecology and human scale[J]. Traffic and Transportation, 2016, 32(2): 34-37. (in Chinese)

[5] CUBELLS J, MIRALLES-GUASCH C, MARQUET O. E-scooter and bike-share route choice and detours: Modelling the influence of built environment and sociodemographic factors[J]. Journal of Transport Geography, 2023, 111: 103664.

[6] PARK Y, AKAR G. Why do bicyclists take detours? A multilevel regression model using smartphone GPS data[J]. Journal of Transport Geography, 2019, 74: 191-200.

[7] CHOU K Y, PAULSEN M, NIELSEN O A, et al. Analysis of cycling accessibility using detour ratios—A large-scale study based on crowdsourced GPS data[J]. Sustainable Cities and Society, 2023, 93: 104500.

[8] SCHNEIDER F, DAAMEN W, HOOGENDOORN S. Trip chaining of bicycle and car commuters: An empirical analysis of detours to secondary activities[J]. Transportmetrica A: Transport Science, 2022, 18(3): 855-878.

[9] ZHANG X L, HUANG J L, BIAN Y, et al. Analysis of factors affecting the severity of accident injuries among shared electric bicycle riders[J]. Journal of Southeast University (English Edition), 2023, 39(3): 284-291.

[10] LUAN S, LI M, LI X, et al. Effects of built environment on bicycle wrong way riding behavior: A data-driven approach[J]. Accident Analysis & Prevention, 2020, 144: 105613.

[11] BIAN Y, YANG J X, ZHAO X H, et al. Study on factors influencing the risky behavior of wrong-way riding of shared electric bicycles based on trajectory data[J]. China Journal of Highway and Transport, 2021, 34(12): 262-275. (in Chinese)

[12] ZHU C H, LI Y, SUN X L, et al. Urban public bicycle demand forecast considering land use[J]. Journal of South China University of Technology (Natural Science Edition), 2022, 50(3): 9-20, 37. (in Chinese)

[13] YUAN J, ZHENG Y, ZHANG C Y, et al. An interactive-voting based map matching algorithm[C]// 2010 Eleventh International Conference on Mobile Data Management. Kansas City, MO, USA, 2010: 43-52.

[14] ZHANG X L, BIAN Y, ZHAO X H, et al. Analysis of factors affecting the risk of overspeeding on shared electric bicycles[J]. Journal of Southeast University (Natural Science Edition), 2024, 54(1): 214-223. (in Chinese)

[15] WANG Y F, DONG Z J, CHEN X D. Research on right-turn stopping detection method for large vehicles in traffic scenarios[J]. Computer Engineering and Applications, 2025, 61(1): 352-360. (in Chinese)

[16] DIXIT M, CHOWDHURY S, CATS O, et al. Examining circuity of urban transit networks from an equity perspective[J]. Journal of Transport Geography, 2021, 91: 102980.

[17] CHAWLA N V, BOWYER K W, HALL L O, et al. SMOTE: Synthetic minority over-sampling technique[J]. Journal of Artificial Intelligence Research, 2002, 16: 321-357.

[18] JELEN G, PODOBNIK V, BABIC J. Contextual prediction of parking spot availability: A step towards sustainable parking[J]. Journal of Cleaner Production, 2021, 312: 127684.

[19] ZHANG D X, LUO T, ZENG Z Y. An improved SMOTE sampling Catboost classification algorithm[J]. Information and Communication, 2020, 33(1): 57-60. (in Chinese)

[20] DING H L, WANG R Q, CHEN T T, et al. A hybrid approach for modeling bicycle crash frequencies: Integrating random forest based SHAP model with random parameter negative binomial regression model[J]. Accident Analysis & Prevention, 2024, 208: 107778.

Funding

National Natural Science Foundation of China(52072012)