An adaptive transportation prediction model for the informal public transport sector in Africa
IEEE ITSC, 8-11 Oct. 2014, Qingdao, China
Intelligent Transportation Systems (ITSC), 2014 IEEE 17th International Conference
The original publication is available at http://ieeexplore.ieee.org.ez.sun.ac.za/xpl/articleDetails.jsp?arnumber=6958102
The informal public transport sector in Sub-Saharan Africa is responsible for transporting the overwhelming majority of the workforce. Often, passengers have to wait for hours for taxis to coincidentally pass by to pick them up, making the transport mode notoriously inefficient. Despite its relevance and impact, the sector is afforded little attention in terms of regulation, development and organization, giving rise to a complex and inefficient system that affects millions of people. In fact, little is known about the industry. To advance understanding of this system, minibus taxis were equipped with tracking devices in this study. Tracking data was then used to develop a model that describes the transport network – essentially finding patterns in the apparent chaos for the potential benefit of its users. The adaptive model uses unsupervised learning to predict the informal stages in the city and provide travelers with intelligence on the best time and place to get transport, thereby reducing the waiting time at the taxi rank and the informal roadside stops. Experimental results show 70.4% model accuracy in dynamically learning the taxi behavior and accurately predicting the best places to get taxis at a given time of the day.