Trip Matrix Development Using Mobile Phone Data: What Have We Learnt So Far and What Are the Key Principles to Follow?
Reza Tolouei, AECOM
this paper proposes an analytical framework which sets out a high level, step by step approach to develop highway trip matrices sourced from mobile data.
Historically, travel demand matrices have been developed in the UK based on collected Origin-Destination (OD) trip information from a sample of whole population (e.g. travel diary data, roadside interview data, etc.), merged with synthetic methods to infill ‘unobserved’ subset of trips or help improve the accuracy of ‘statistically unreliable’ trip estimates within the matrix. More recently (since 3 to 4 years ago), mobile phone positioning data (referred to as ‘mobile data’) have been used increasingly by the transport planning community in the UK as a replacement to the above data sources in developing demand matrices.
In using conventional data sources to develop OD matrices for highway assignment, general guidance and established ‘best practice’ is available in the UK, mainly sourced from Design Manual for Roads and Bridges (DMRB) and leading transport modelling experts within the industry. These set out the key principles that should be followed in developing trip matrices using conventional data sources. However, such guidance or established practice is not readily available to provide essential support to transport modellers in developing trip matrices based on mobile data.
It has been made clear by a number of recent studies that there are fundamental differences in the nature of the data and their errors and biases, and therefore in the challenges faced, when mobile data is used as the primary source of trip matrix development. A recent paper provided the key considerations for matrix development and presented some of the lessons learnt regarding use of mobile data, based on the experience gained by the Technical Consistency Group established by Highways England to develop highway ‘prior’ matrices for Regional Traffic Models (Khorgami, et. al., 2016). Two other recent papers discussed an approach to verify and use mobile data, and more importantly, quantified evidence on how the matrices are compared with those developed based on conventional methods (Tolouei, et al., 2015, and Tolouei, et al., 2016).
Building on the above 3 papers, as well as several other matrix development studies directly led by the author or undertaken elsewhere, this paper proposes an analytical framework which sets out a high level, step by step approach to develop highway trip matrices sourced from mobile data.
The paper brings together lessons learnt so far, the key challenges that the industry is facing in using the data, and the sequential steps that could be taken to address these in order to fulfil the key requirements of trip matrices. It should be clarified that it is not the intention of this paper to provide any specific matrix development guidance; it is however intended to set out key principles, mainly drawing on existing knowledge, evidence, and experience. A particular emphasis is given to various sources of bias in the data, and ways to address these in a systematic and pragmatic manner.
Whilst the focus of the proposed analytical framework is the post-processing of the matrices produced by Mobile Network Operators (MNOs), the paper includes a discussion of the existing gaps between the data analyses undertaken by the MNOs, and the requirements by transport models, and how these could be bridged.
The paper also discusses how relevant the existing matrix development principles and testing criteria are, when mobile data is the primary source of data instead of conventional data sources. Finally, key areas of development are identified, with proposed short term, medium term, and long term measures to achieve these.
Tolouei, R., Psarras, S., Prince, R. (2016). “Origin-Destination Trip Matrix Development: Conventional Methods vs. Use of Mobile Phone Data”, proceedings of the European Transport Conference, Barcelona, 2016.
Khorgami, S., Powell, G., Tolouei, R., Hanson, P. ”Use of Mobile Phone Data in Regional Traffic Models: Experience from Highways England”, Transport Practitioners Meeting, Nottingham., 2016.
Tolouei, R; Alvarez, P; Duduta, N. “Developing and Verifying Origin-Destination Matrices Using Mobile Phone Data: the LLITM case.” European Transport Conference, Frankfurt., 2015.
Association for European Transport