Abstract

Mobile network data (MND) are an increasingly important data source for official statistics, e.g. for population statistics, mobility analysis or everyday policy decisions in general. To this end, Germany’s Federal Statistical Office (Destatis) is working with universities and mobile network operators (MNO) towards developing new solutions to address the qualitative challenges of this data. In this presentation, building on work of the past years and its vast experience in the field, Destatis will focus on using MND in the field of traffic and mobility planning and present recent research findings as well as its current research strategy in the field.
One present goal is to make extrapolation methods for MND more efficient, e.g. to map traffic networks more precisely. Here, statistical simulation is used to investigate how the consistency of complex traffic forecasts can be improved. By doing so, Destatis is aiming at broadening the existing database in Germany, one that examines the use of mobile telephony in detail and, together with this, provides better information on the socio-demographic background of mobile phone users. This data, based on a large-scale survey, is to be linked with mobile phone data from two network operators, helping to compensate for existing limitations in the quality of MND.
Destatis’ work roots in the belief that high-quality decisions can only be obtained to a limited extent with conventional data. MND allow for significantly expanded insights, especially in a very up-to-date, small-scale and dynamic form. Initial work with MND has shown that selectivity is a major challenge, which is why there is a clear need for expanded research in this area before this data source can be used in a high-quality manner. To this end, new digital data are combined with traditional data. The aim of current research at Destatis is to develop procedures for the statistically valid use of this data source.
Building on the results of past projects, the known quality limitations of mobility data are to be compensated as best as possible by a survey tailored to these questions in accordance with the Federal Statistics Act and statistical methods such as small area models. Finally, qualitatively optimized origin-destination matrices, based on survey, administrative and mobile phone data, will be used on a pilot basis for selected regions to forecast mobility scenarios within the framework of microsimulations.