Improving Service Quality of Metro Systems—A Case Study in the Beijing Metro

In this study, we propose a method that combines Bayesian network, structural equation modeling, and importance-performance analyses to evaluate and improve the service quality of crowded metros from the point of service components. First, service components that influence the service satisfaction of crowded metros are investigated, and 7 service components and 27 factors are collected from the perspective of metro daily operation. Correspondingly, 8,011 sample questioners are collected using both online (i.e. microblog and WeChat) and face-to-face survey methods. The usage of these service components is more practical for metro managers to improve service quality than the factor analysis methods used in previous studies. Second, a hybrid method combining Bayesian network and structural equation modeling is proposed to establish the relationship between service quality and these service components and factors. This hybrid method can learn the uncertain relationship directly from the data, explain the psychological element, and pinpoint the crucial service component. It is concluded that security check and waiting for boarding have the greatest effect on the overall satisfaction. With the help of importance-performance analyses, some practical suggestions (e.g., theme stations, women-only carriages, and one-day ticket) are proposed to optimize bottleneck service components and thus improve the service quality of daily metro operation. Further, heterogeneity of service quality satisfaction for various groups of respondents is discussed, which contributes to effective strategies for satisfaction improvement.