Activities of the GWG task team on Mobile Phone Data

Mobile phones are used by large segments of the population in all parts of the world, and it is thus expected that mobile phone data could fill data gaps worldwide. In its 2018 "Measuring the Information Society Report", ITU showed that the average mobile subscription rate is 107 per 100 inhabitants worldwide, with a lower average in Africa (76 per 100). These numbers show how pervasive mobile phone use is. Almost every person in the world lives within reach of a mobile-cellular signal.

Mobile phone data could help determine where tourists and migrants come from, how long they stay and where they go. The granularity of information which potentially can be obtained from mobile phone data is much higher than what can be obtained through traditional surveys. Moreover, the time lag from data collection to analysis could also be significantly reduced.

The GWG task team on the use of mobile phone data for official statistics conducts use cases, develops methodology and runs training workshops to build capacity in various regions of the world. It delivers on project activity, training workshops and preparation of training materials. Details of these activities are given hereafter.

Projects

International Meeting on Measuring Human Mobility, 27-29 March 2019, Tbilisi, Georgia

A project was set up in Georgia on measuring human mobility (as part of the deliverables of the GWG task team) which aims to estimate population mobility patterns broken down by migrants, seasonal workers and tourists. The international meeting in Tbilisi in March this year was built on three parts, namely (1) measuring human mobility using mobile phone data, (2) compiling migration and tourism statistics using traditional data sources, and (3) project implementation using the UN Global Planform.

The national statistical office in Georgia (Geostat), the Georgian mobile phone regulator (GNCC) and members of the task team (ITU, UNSD, Eurostat, Positium and others) work together to develop and test methods to estimate migration and tourism statistics in Georgia with use of mobile phone data. Thereafter, other countries can test and validate these new methods, such as Indonesia, Malaysia, Oman, Italy, the Netherlands, Columbia and Egypt.

In addition to the Georgia project, new projects are being set up in Malaysia, Thailand, Timor Leste and the Pacific Islands with support of BPS Indonesia, UNESCAP, Positium and UN Global Pulse.

Workshops

Training workshop at the 5th international conference on Big Data for official statistics in Kigali, Rwanda, 29 April - 1 May 2019

Under the heading of unlocking the potential of big data analytics, a mobile phone data workshop was held in Kigali, Rwanda. The workshop was attended by about 25 participants from 12 different countries. These mobile phone data could offer unique opportunities for national statistical offices in Africa. However, you need to acquire knowledge and skills to apply new methods, new algorithms and use new data processing tools. Other important issues which were addressed were privacy and regulatory concerns, data access models, and partnership models. Indonesia demonstrated how they used mobile positioning data for measuring domestic tourism, underlining the benefits and challenges. Mr Siim Esko from Positium presented on how to design, prepare and build a data project using mobile phone data.

Regional workshop on the use of mobile phone data for official statistics, 11-14 June 2019, Jakarta, Indonesia

Similar to the workshop in Kigali, this workshop highlighted the use of mobile phone data in the statistical production process. Again about 25 countries from 10 different countries participated. The workshop explained what mobile phone data are, why and how they need to be pre-processed and how we can derive statistics from them. How mobile phone data can complement traditional data sources and the use of non-traditional methods and data sources for producing official statistics were also discussed. Sample data sets were used for hands-on work. Further, case studies were described from a few countries where mobile phone data have already been used for official statistics. The results of this workshop also fed into the development of an e-learning training programme on use of mobile phone data for official statistics.

UNESCAP Statistics Brief June 2019

The Statistical Offices of Bangladesh, Cambodia, Georgia, Indonesia, Malaysia, Mongolia, Nepal, the Philippines, Thailand and Viet Nam, ten in all, met in Jakarta, Indonesia, 11-14th June 2019 to share experiences on why and how to use mobile phone data for official statistics. The use of mobile phone data in Asia and the Pacific can be considered as in its infancy but interest is strong.

Positium, an Estonian based tech company, shared its country’s experience in why and how to use mobile phone signaling data. In the case of Estonia, the data is applied to measure various statistics, including population, tourism and human mobility statistics. In Asia and the Pacific, BPS Indonesia has started using signaling data from mobile phones in the production of its cross-border inward tourism statistics for those cross-border areas where there is no immigration office collecting data. This work was then expanded into all cross-border inward tourism to validate official statistics produced using traditional data collection methods. These experiences highlight how mixed modes can improve statistics quality substantially. BPS Indonesia are looking to expand to other types of tourism statistics such as domestic tourism, particularly as their confidence in the tourism statistics based on signaling data grows.

Over the course of the four-day workshop, countries shared their thoughts about the potential benefits of using mobile phone data in their own context. Ten key reasons were identified:

• Much more detailed statistics can be obtained through sub-national estimates and granularity of data
• Mobile phone data can help validating sampling frames
• They can also help validating survey statistics
• They can improve accuracy
• ... while reducing cost
• They can help improving timeliness
• ... while reducing respondent burden
• They can meet unmet demand by creating new kinds of statistics, which were just not possible to produce with traditional methods
• Importantly, they can help fill data gaps for the SDG indicators
• And finally, they can help building new business models.

Read the full brief.

Training Materials

Handbook on use of Mobile Phone data for official statistics

The handbook consists of four main parts: applications, explanation of the data source, methodology and dissemination. Applications are shown for producing tourism and event statistics, population statistics, migration statistics, commuting statistics and traffic flow statistics. For example, a study done by the Italian statistical office (ISTAT) focused on 39 municipalities in the province Pisa, Tuscany. The dataset consisted of 7.8 million CDRs collected in the month of January 2012 for more than 200,000 users with a national mobile phone contract (no roaming users were included). The results showed inter-city mobility among these municipalities and results also showed that flow estimates obtained with the so-called Sociometer method were more accurate for larger towns than for smaller towns.

The data sources chapter provides explanation about various technologies used in generating those data sources. For example, wireless technologies (such as 2G, 3G and 4G) that are used in mobile telephones and telecom networks worldwide follow different standards and specifications composed and maintained by organizations such as ETSI, 3GPP, Qualcomm and IEEE. In Europe (and also Japan and China), the most prevalent mobile communication technologies today are GSM (2G) and UMTS (3G), which follow the 3rd Generation Partnership Project’s (3GPP27) technical specifications. For this reason, most of the descriptions of mobile network operators’ (MNOs) network architecture and data access from the MNOs’ systems in this handbook will be based on 3GPP specifications.

E-learning course

On the basis of the handbook the task team is developing an e-learning course. Text is sourced from the handbook, whereas video is taken from the actual training workshops, and exercises are mixed in from the exercises used at those workshops.

Synthetic data set

Our partners of ONS Data Science Campus and Positium have started to develop a synthetic data set of mobile phone records. Such data set shows behavior of real mobile phone records, but do not contain any personal sensitive data. Such dataset is therefore perfect to develop and test algorithms in a fully cooperative setting.

Other training materials

The training materials have been made available for public use on the GWG website. It contains various PowerPoint presentations.

Partnerships

The task team can only achieve results if it can count on the participation of all its partners. Besides the national statistical office and international agencies, work in this area needs strong support from private sector, academia and civil society, such as Positium, Flowminder, UN Global Pulse, Telenor Research and GSMA.

Ronald Jansen

Ronald Jansen is Assistant Director of the United Nations Statistics Division in New York. He is responsible for the Division's work on data innovation and on capacity development strategies. This includes leading the intergovernmental processes for innovation of official statistics and implementing the capacity development pillar of the Cape Town Global Action Plan in support of the monitoring of the 2030 agenda for sustainable development.