Cluster Analysis for Diminishing Heterogeneous Opinions of Service Quality Public Transport Passengers

Abstract

[EN] One of the principal measures that public transport administrations are following for reaching a sustainable transportation in the cities consists on attract a higher number of citizens towards the use of public transport modes, by offering high quality services. Collecting users opinions is the best way of detecting where the service is failing and which aspects are been provided successfully. The main problem that has to be faced for analyzing service quality is the subjective nature of its measurement, offering heterogeneous assessments among passengers about the service. Stratifying the sample of users on segments of passengers which have more uniform opinions about the service can help to reduce this heterogeneity. This stratification usually is conducted based on the social and demographic characteristics of the passengers. However, there are more advance techniques that permits to identify more homogeneous groups of users. One of these techniques is the Cluster Analysis, which is a data mining technique that can be used for segmenting the sample of passengers on groups that share some common characteristics, and that have more homogeneous perceptions about the service. This technique has been applied in other fields of transport engineering but it has never been applied for searching homogeneous groups of users with regards to service quality evaluation in a public transport service. For this reason, the aim of this work is to find groups of passengers that perceive the quality of the service in a more homogeneous way, and to apply to this clusters a suitable statistic technique that permit us to discover which are the variables that more influence the passengers¿ overall evaluation about the service. The comparison among the results of each cluster will show considerable differences among them and also with the results obtained using the global sample.