The integration of informal minibus-taxi transport services into formal public transport planning and operations - A data driven approach

Master Thesis


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title

University of Cape Town

The MiniBus Taxi (MBT) mode is poorly understood by planning and operational authorities, yet plays a big role in the economies of developing countries transporting the workforce to and from their places of employment and offering employment to thousands in the operations of these services, as well as the numerous rank-side services and amenities offered to patrons. In recent years, research focussed on mapping paratransit services, including MBTs, in cities of the developing world has contributed significantly to the understanding of the mode in terms of its spatial extent in its respective service areas. In South Africa, experience has shown that the wholesale replacement of MBTs with scheduled services is an unattainable goal. Instead, planning authorities and researchers have, more recently, shown interest in investigating feasible methods of integrating the scheduled and unscheduled services as hybrid planned-trunk and paratransit-feeder networks. The objective of this research is to present the case for simple methods of planning and carrying out onboard surveys of paratransit services to classify and to better understand the operations of individual routes, identified route classes, the network as a whole, as well as revealed passenger demand for the services and, ultimately, how this information can be wielded in the planning and implementation of hybrid routes or networks. The data central to this study consist of onboard captured MBT data, which was collected with a public transport data capturing application using GPS enabled smartphones in the City of Cape Town from April to August 2017 as part of a City of Cape Town’s Transport and Urban Development Authority (TDA) data collection project. The purpose of the project was to clarify the actual extent of MBT services within the City and to improve the representation of the MBT mode in the City of Cape Town’s travel demand model. An Android smartphone application, purpose-built for collecting operational information onboard public transport vehicles, was used to collect spatial and temporal data on the operations of a sample of active MBT routes in Cape Town. The application, which saw some functionality updates specifically for the project, was used to collect the following information per MBT trip: · Location of stops; · Time of arrival and departure at stops; · Number of passengers boarding and alighting at each stop; · The relative boarding and alighting stop of each specific passenger; · The amount paid in fare money per passenger at each stop; · The actual path travelled by the vehicle as a GPS route trace; and · The origin and destination route description of each route captured. It is estimated that there are more than 800 active and operational routes in the Cape Town. The objective of the data collection project was to survey each one of these routes for a prespecified number of trips. As the project was still underway when this research was carried out, the information listed above collected for a sample of trips for 278 routes (556 if the reverse direction is considered as a unique route designation) formed the basis of this study. During the course of this study, the analyses of these data have shown that while the operational characteristics of individual routes are relatively consistent and stable, it is possible to distinguish between different service typologies within the larger route network. From the raw data structure listed above, the operational characteristics that were calculated for each trip and aggregated at the route level included: · Trip and route distances; · Average operating speeds; · Travel times; · Number of stops per trip; · Load factors between stops along the route; and · Fare rates and trip revenues. In addition to the identification of the operational characteristics of the MBT network, service classes and routes, the outcomes of the study include providing a framework of methods for the collection, extraction, cleansing, analysis and visualisation of the data. It also includes the identification of metrics which are key in describing the difference in service types. The descriptive operational characteristics that were calculated for each trip record, inbound and outbound per route, were evaluated to establish whether they can be used to determine if different service typologies can be observed in the data. It was found that simple k-means clustering procedures may be used to classify the routes into separate, distinguishable service classes. For the purpose of this study, it was decided, nominally, that the classification should be executed for three classes. Three was subjectively considered a good value to be inclusive of traditional Trunk and Feeder or Distribution, route types as well as the possibility of the existence of a yet to be defined third type. The clustering procedures were carried out for different combinations of the operational variables for which the most consistent results were obtained for the combination distance – stop density1 – passenger turnover. Analysis of the within-class operational characteristics indicates that these three service classes clearly differ in terms of their stop frequencies, distances, speeds and their spatial network coverage. The study furthermore provides evidence that the understanding of the MBT network and sub-networks of service classes within this network, including its interaction with other public transport modes and infrastructure, provides planning and operating authorities with key information for effectively planning and implementing hybrid networks. Finally, the study demonstrates many additional insights can be garnered from these data by implementing improved statistical sampling and survey methods at the route level and by analysing aspects of the data that were not considered central to the research. These aspects include route adherence studies, origin – destination studies and methods of expanding the onboard data samples accurately by marrying it with data collected during static rank departure and arrival counts. Ultimately, the study shows that an unprecedented knowledge of the operations of MBT routes and networks may be obtained through detailed yet simple analysis of onboard data and that this knowledge may be very useful in the planning and operations of integrated public transport networks.