A rise in population, along with urbanization, has been causing an increase in demand for urban transportation services in the sub-Saharan Africa countries. In these countries, mobility of people is mainly ensured by bus services and a large-scale informal public transport service, which is known as paratransit (e.g., car rapides in Senegal, Tro Tros in Ghana, taxis in Uganda and Ethiopia, and Matatus in Kenya). Transport demand estimation is a challenging task, particularly in developing countries, mainly due to its expensive and time-consuming data collection requirements. Without accurate demand estimation, it is difficult for transport operators to provide their services and make other important decisions. In this paper, we present a methodology to estimate passenger demand for public transport services using cell phone data. Significant origins and destinations of inhabitants are extracted and used to build origin-destination matrices that resemble travel demand. Based on the inferred travel demand, we are able to reasonably suggest strategic locations for public transport services such as paratransit and taxi stands, as well as new transit routes. The outcome of this study can be useful for the development of policies that can potentially help fulfill the mobility needs of city inhabitants.

Transport-on-demand services, such as demand-responsive transport (DRT), involve a flexible transportation service that offers convenient and personalised mobility choices for public transport users. Integrating DRT with mobile data and data analytics enhances understanding of travel patterns and allows the development of improved algorithms to support design-optimised services. This study introduces a replicable framework for DRT that employs an on-demand transport simulator and routing algorithm. This framework is supported by a mobile data set, enabling a more accurate service design grounded on actual demand data. Decision-makers can use this framework to understand traffic patterns better and test a DRT solution before implementing it in the actual world. A case study was conducted in Porto, Portugal, to demonstrate its practicality and proof of concept. Results show that the DRT solution required 135% fewer stops and travelled 81% fewer kilometres than the existing fixed-line service. Findings highlight the potential of this data-driven framework for urban public transportation systems to improve key performance metrics in required buses, energy consumption, travelled distance, and stop frequency, all while maintaining the number of served passengers. Under specific circumstances, embracing this approach can offer a more efficient, user-centric, and environmentally sustainable urban transportation service.

Simulation modeling is a well-known and recurrent approach to study the performance of urban systems. Taking into account the recent and continuous transformations within increasingly complex and multidimensional cities, the use of simulation tools is, in many cases, the only feasible and reliable approach to analyze such dynamic systems. However, simulation models can become very time consuming when detailed input-space exploration is needed. To tackle this problem, simulation metamodels are often used to approximate the simulators' results. In this paper, we propose an active learning algorithm based on the Gaussian process (GP) framework that gathers the most informative simulation data points in batches, according to both their predictive variances and to the relative distance between them. This allows us to explore the simulators' input space with fewer data points and in parallel, and thus in a more efficient way, while avoiding computationally expensive simulation runs in the process. We take advantage of the closeness notion encoded into the GP to select batches of points in such a way that they do not belong to the same high-variance neighborhoods. In addition, we also suggest two simple and practical user-defined stopping criteria so that the iterative learning procedure can be fully automated. We illustrate this methodology using three experimental settings. The results show that the proposed methodology is able to improve the exploration efficiency of the simulation input space in comparison with non-restricted batch-mode active learning procedures.