EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Ordinal regression and explainable machine learning reveal key determinants of urban bus service quality in Dhaka

Joydeep Banik, Arman Hossain, Md Sifat Bin Siraj, Md Jubayadul Islam, Zahirul Haque Molla, Mehedi Hasan, Armana Sabiha Huq

Discover Cities · 29.07.2026 · 3 (1)

DOI: 10.1007/s44327-026-00334-9 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Discover Cities, online erschienen am 29.07.2026. Verfasst von Joydeep Banik und 6 weiteren Autor:innen. Die Arbeit nennt 56 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Bangladesh is a developing country experiencing rapid population growth, necessitating improved urban transportation. Buses could be a viable solution, but a trend towards personal vehicles persists. To understand this shift, the quality of current bus services needs to be evaluated. Improving bus service quality promotes sustainable urban mobility by reducing congestion and car reliance in rapidly developing urban centres. Therefore, this study assesses the service quality of urban buses in Bangladesh and identifies key dimensions that should be prioritized to enhance overall bus quality. A SERVQUAL analysis was conducted using data collected from a questionnaire survey on five key bus routes in Dhaka city. A total of 500 respondents were interviewed to assess their perceptions and expectations of current services. Ordinal Logistic Regression (OLR) was applied, along with machine learning models such as Extreme Gradient Boosting (XGBoost) and Random Forest (RF). The RF performed better in the satisfaction model with an accuracy of 0.71, F-1 Score of 0.73 and AUROC (Macro) of 0.82. In contrast, the XGBoost performed better in the dissatisfaction model with an accuracy of 0.70, F1-Score of 0.69 and AUROC of 0.79. OLR showed a pseudo R 2 of 0.199 for satisfaction and 0.182 for dissatisfaction model. After analyzing the gap scores between perception and expectation, the Level of Service (LoS) was categorized as Category 3. SHAP plots using Shapley values identified key SERVQUAL dimensions. Major service gaps include bus schedules, maintenance, ticketing, safety, and professionalism. Dissatisfaction mainly stems from a lack of assurance and empathy, such as safety concerns, indifference, and unprofessionalism. Improving assurance, reliability and responsiveness such as professionalism, safety, prompt help and maintenance, can boost customer satisfaction.

Zitieren

APA

Banik, J., Hossain, A., Siraj, M. S. B., Islam, M. J., Molla, Z. H., Hasan, M., & Huq, A. S. (2026). Ordinal regression and explainable machine learning reveal key determinants of urban bus service quality in Dhaka. Discover Cities, 3(1). https://doi.org/10.1007/s44327-026-00334-9

Vancouver

Banik J, Hossain A, Siraj MSB, Islam MJ, Molla ZH, Hasan M, et al. Ordinal regression and explainable machine learning reveal key determinants of urban bus service quality in Dhaka. Discover Cities. 2026;3(1). doi:10.1007/s44327-026-00334-9

BibTeX
@article{Banik2026,
  author = {Banik, Joydeep and Hossain, Arman and Siraj, Md Sifat Bin and Islam, Md Jubayadul and Molla, Zahirul Haque and Hasan, Mehedi and Huq, Armana Sabiha},
  title = {Ordinal regression and explainable machine learning reveal key determinants of urban bus service quality in Dhaka},
  journal = {Discover Cities},
  year = {2026},
  volume = {3},
  number = {1},
  doi = {10.1007/s44327-026-00334-9}
}
Kostenlos starten →