Abstract:
This thesis uses a multidimensional analytical approach to provide a complete evaluation of the service quality (SQ) of Dhaka's Mass Rapid Transit (MRT) Line-6. It specifically focuses on understanding and quantifying the attributes influencing the perceived SQ of this recently implemented urban transit system.
To achieve the study objectives a thorough examination of existing research on Public Transport (PT) service quality assessments, highlighting various methodologies and attributes has been conducted to evaluate overall SQ of MRT.
Considerable emphasis is placed on the Structural Equation Modeling (SEM) methodology, which has been extensively adopted within the realm of the transportation industry. SEM is identified as the principal analytical instrument, supplemented by a suite of additional methodologies including Probit Linear Regression, Ordinal Linear Regression, Multinomial Linear Regression, and various Machine Learning Algorithms such as the Random Forest Classifier, Support Vector Machine, and Decision Tree-based frameworks like CART and C4.5, for the comprehensive evaluation of service quality (SQ). This combination of methodologies results in a rigorous study, with SEM appearing as the best method of assessing the complex relationships of SQ attributes and Random Forest Classifier for ranking the influence of these attributes on overall SQ in the context of MRT of Dhaka city.
The proposed methodology consists of a three-step research process, including data gathering through a specially designed questionnaire survey, selection of SQ variables through extensive literature review and professional consultations and development of structural equation models to understand the relationships between SQ variables. A total of twenty-nine (29) SQ attributes and three (3) Latent Constructs were chosen for the research, classified as exogenous, endogenous, or latent variables, and the SEM approach was used to model these variables.
Preliminary analysis of the data focuses on descriptive insights into the demographic characteristics, income, and occupation of respondents. The survey results indicate a high level of comfort and reliability as the primary reasons for choosing MRT, with the majority of users previously relying on buses (60%), Auto-rickshaws (15%) and Motor-cycle (7%) for their mode of transportation. During the analysis of different models for SQ assessment, this thesis confirms the superiority of SEM in handling the complexities of inter-relationship among the attributes having a Comparative Fit Index (CFI) value of 0.72, Tucker-Lewis Index (TLI) value of 0.69 and co-efficient of determination (CD) of 0.97 assessing the overall SQ in MRT. Among the latent attributes 'Affordability and Accessibility' impacting the overall SQ of MRT in most significant way and having positive correlation with other two latent constructs named 'Metro Service Efficiency' and 'Passenger Perception and Experience'.
However, the applicability of probit (R2=0.44), ordinal (R2=0.45), and multinomial regression (R2=0.46) has very poor fit indices, indicating that these linear regression approaches are not suited for SQ modelling for MRT in Dhaka or any public transit system as in most cases the responses of users have non-linearity.
Moreover, among the ML algorithms Random Forest Classifier shows an accuracy of 88.21%, whereas, SVM shows an accuracy of 86.07%. CART shows an accuracy of 83.57%, whereas C 4.5 shows an accuracy of 82.86% revealing that the Random Forest Classifier achieves the best fitting model for the perceived SQ of MRT Line-6. Moreover, it reveals ‘Inclusive Service Performance’, ‘Customer Loyalty’, ‘Switching Cost Comparison’, ‘Lighting Near MRT Station’ and ‘MRT Ticket Fare’ are found to contribute positively and significantly to the overall SQ.
Furthermore, this study contributes significantly to urban transportation planning literature, offering practical insights for policymakers, urban planners, and transport authorities in enhancing MRT services in Dhaka. The findings will pave the way for a more efficient, sustainable, and user-friendly metro transit system in Dhaka as mentioned in RSTP 2015. The comprehensive approach and methodology of this research can serve as a valuable reference for future studies in urban transportation and service quality assessment.