the digital economy •Innovation Union: It consists of over 30 action points aimed at boosting research
initiatives of the Europe 2020 Strategy, has the goal of creating a flourishing digital economy by 2020
The available data indicates that while the FDI sector is of vital importance to the growth of exports
Subdivision of data into North & South Tipperary areas not available 28 Case study: Benefi ts of participation in EU Projects
-refining & Bioenergy, Data Analytics and Manufacturing Research. Two Research Centres are hosted by the University of Limerick:
c) Data pertaining to employment, turnover and exports in the Region from the Central Statistics
A national clustering policy is essential to provide support and structure to cluster development. Additionally, there is a requirement for policies to ensure the creation of the type of
development and the regional benefits of clustering •To carry out the necessary prioritisation research and analysis in the Midwest Region to identify
Baseline Data: The initial sections of this Strategy form a baseline assessment of the current research
•Priority Area B-Data Analytics, Management, Security & Privacy •Priority Area C-Digital Platforms, Content & Applications
the lack of up-to-date data, in particular in developing countries. ITU is joining the international statistical community in looking into ways of using new and emerging data sources †such as those
associated with big data †to better provide timely and relevant evidence for policy-making. Calls for
a â€oedata revolution†are prominent in the international debates around the post-2015 development agenda, and ICTS have an important role to play in view of their capacity to produce,
huge amounts of data, as well as being a major source of big data in their own right. Big data from
mobile operators, for example, are real-time and low-cost and have one of the greatest development potentials in view of the widespread use and availability of mobile networks and services.
This report provides the reader with a comprehensive and critical overview of the role of big data from the
telecommunication sector, for use in social and economic development policy and for monitoring the future information society
I trust that the data and analysis contained in this report will be of great value to the ITU membership
The 2014 edition of the Measuring the Information Society Report was prepared by the ICT Data and
and Michael Minges to the compilation of data on international bandwidth, revenue and investment. Helpful inputs and suggestions were received
The report includes data from Eurostat, OECD, IMF, Informa, the UNESCO Institute for Statistics, the
ITU also appreciates the cooperation of countries that have provided data included in this report The report was edited by Anthony Pitt and Bruce Granger, ITU English Translation Section.
Chapter 5. The role of big data for ICT monitoring and for development...173 5. 1 Introduction...
5. 2 Big data sources, trends and analytics...175 5. 3 Telecommunication data and their potential for big data analytics...
181 5. 4 Big data from mobile telecommunications for development and for better monitoring...185 5. 5 Challenges and the way forward...
195 Chapter 5 Annex...207 List of references...213 Annex 1. CT Development Index (IDI) methodology...
Annex 2. ICT price data methodology...231 Annex 3. Statistical tables of indicators used to compute the IDI...
5. 1 The five Vs of big data...176 5. 2 An overview of telecom network data...
182 5. 3 Customer profiling using telecom big data...184 xii List of boxes 1. 1 Final review of the WSIS targets:
Achievements, challenges and the way forward...26 1. 2 A decade of successful international cooperation on ICT measurement...
1. 4 What is a data revolution?..30 2. 1 ITU discussion forums on ICT statistics...
5. 1 How big data saves energy †Vestas Wind Systems improves turbine performance...177 5. 2 How Twitter helps understand key post-2015 development concerns...
5. 3 How mobile operators currently use data to track service uptake, business performance and
5. 4 Using mobile data for development...187 5. 5 How mobile network data can track population displacements †an example from the
2010 Haiti earthquake...188 5. 6 Leveraging mobile network data for transportation and urban planning in Sri lanka...
189 5. 7 Poverty mapping in CÃ'te d†Ivoire using mobile network data...190 5. 8 Using mobile-phone data to track the creditworthiness of the unbanked...
191 5. 9 Using mobile big data and mobile networks for implementing surveys...193 List of tables
1. 1 Rural population covered by a mobile-cellular signal, 2012.4 1. 2 Total Internet domain registrations by world region, 2003,2008 and 2013.18
2. 1 IDI values and changes, 2012 and 2013.41 2. 2 ICT Development Index (IDI), 2012 and 2013.42
5. 1 Sources of big data...175 1 Measuring the Information Society Report 2014 Chapter 1. Recent information society
The data also show a continuous increase in Internet usage, with growth in the number of
and data) market segments, and considering both subscriptions and household access data. This will be followed by a presentation of the latest
trends in terms of investment and revenue in the telecom sector. Then, a number of key indicators
WSIS+10 review, the demand for a data revolution and the role of big data for ICT monitoring
1. 2 The voice market In line with developments in recent years, fixed telephony is on the decline in all regions of the
subscription or SIM CARD †the total numbers and growth rates strongly point to market saturation Whether this will change in the near future, and
ITU World Telecommunication/ICT Indicators database 39.2 26.3 24.9 15.8 12.7 8. 7 1. 3
data reveal, however, that digital divides still exist and that some people are excluded still from access to communication networks
ITU World Telecommunication/ICT Indicators database 162.7 124.7 109.9 108.5 96.4 89.2 69.3 0 20
For countries where data are available, the number of mobile subscriptions far exceeds the number of mobile phone users (Partnership, 2014
many developing countries for which data are available, but is closing with the availability of
Further research and data would be necessary to determine people†s access to, and use of, voice
Partnership (2014) based on ITU data Overall mobile-cellular population coverage %Rural population covered %Rural population
ITU World Telecommunication/ICT Indicators database 27.7 16.7 14.3 9. 8 7. 7 3. 1
The data on fixed-and mobile-broadband uptake confirm what has been observed on the ground. In developed countries, fixed
ITU World Telecommunication/ICT Indicators database 83.7 32.0 21.1 6. 3 0 10 20 30
smartphone, tablet) and SIM CARDS Looking towards the future, the growth potential for mobile broadband looks promising, as
data based on ITU and Telecom Advisory Services calculations more and more countries upgrade their mobile networks. As mentioned earlier, 2g population
Data on 3g population coverage are less available. According to ITU estimates, global 3g population coverage stood at around 50 per cent
New data collected by ITU on the deployment of fibre transmission capacity in countries shows
at the data also reveals major disparities across regions: Asia and the Pacific (in particular China
Partnership (2014) based on ITU data Percentage of rural population covered by at least a 3g mobile
delivering data-intensive applications and services through high-speed networks. While fibre transmission networks constitute an
ITU World Telecommunication/ICT Indicators database connectivity, because of the strong internal demand and also its location:
ITU World Telecommunication/ICT Indicators database 221 420 1'213 702 4'384 11'572
The latest ITU data show that by end 2014, almost 44 per cent of the world†s households will have
ITU World Telecommunication/ICT Indicators database 78.0 57.4 53.0 43.6 36.0 35.9 11.1 0 10
data are available, rural household access falls far below urban household access, with differences ranging from 4 per cent (meaning
2014). 8 Available data also show that Internet access in rural households is growing slowly
but data are not readily available for those countries. As has been illustrated earlier, network deployment is still
Data on public access is collected by ITU through its household questionnaire, but only few countries
report data on this indicator, let alone broken down by urban and rural populations. Public access can be provided by commercial facilities
ITU World Telecommunication/ICT Indicators database 78.3 40.4 32.4 8. 0 %0 10 20 30
The data include both global top-level domain (gtld) and country code top-level domain (cctld) registrations
Data sourced from Facebook 0 500 1'000 1'500 2013 2012 2011 2010 2009
International data on ICT access and use by enterprises are collected annually by the United Nations Conference on Trade and Development
In the developing world, data on ICTS in enterprises are scarce and only collected by few
Data supplied by Zooknic, compiled from cctld and other sources. Figures exclude fifteen cctlds which act as virtual
There is little data on the use of ICTS by government organizations, and those countries that do have data are usually the more advanced
ones with high levels of connectivity in general More information is available about government services provided online, tracked by the United
The latest data show that, today, governments of all countries have established central websites and that more
UNCTAD Information Economy Database, 2014, available at unctadstat. unctad. org P e rc e n
Data from United nations E-government Survey (2014 Chart 1. 21: E-government services provided by countries (transactional services, left, and e-participation
Data from United nations E-government Survey (2014 101 73 60 46 44 42 41 40
exchange of forms, data and other information Internet access in schools also achieves cost efficiencies by automating manual tasks and
The latest available data from the UNESCO Institute for Statistics (UIS) 22 show that, in
developed countries for which data are available In developing countries, school access to Internet is lower on average, although much
Data on broadband in secondary schools in Bangladesh are not available. Data for Nicaragua, Philippines and Indonesia do not include upper
secondary. Data for European countries and Costa rica refer to lower secondary. Data for Guyana, Nicaragua and Indonesia refer to primary and lower
secondary. Data for Cambodia include pre-primary schools. Data for Morocco, Tunisia, Guyana, Montserrat, Dominican republic, Nicaragua, Colombia
Trinidad and tobago, Bangladesh, Philippines, Sri lanka, Azerbaijan, Bhutan, Cambodia, Kazakhstan, Malaysia, Maldives, Singapore, Belarus and the Russian Federation refer to public schools.
In Suriname, there are no private schools in upper secondary. Data for Palestine refer to West bank schools
only Source: UIS database, Partnership on Measuring ICT for Development WSIS Targets Questionnaire, 2013 P
e rc e n ta g e o f s c h o o ls
Americas 0 10 20 30 40 50 60 70 80 90 100 Internet Fixed broadband Internet
Data for Philippines refer to primary and lower secondary. Data for Venezuela refer to primary only.
Data for Montserrat and Saint lucia refer to secondary only. Data for Palestine refer to West bank schools only.
Data for Bahrain, Belarus, Morocco and Tunisia refer to 2008. For Morocco, ICT -qualified teachers figures refer to 2008.
Data for Azerbaijan, Barbados, Jordan, Saint lucia, Singapore, Trinidad and tobago, Uruguay, Philippines and Sri Lanka refer to public schools only
Source: UIS database, Partnership Questionnaire on WSIS Indicators, 2014 Proportion of teachers trained to teach basic computer skills (or computing
%P ro p o rt io n o f te a c h e rs
t ra in e d t o t e a c h s u b
Available data collected by UIS at the international level shows that education systems in countries seem to put more emphasis on
document also includes two targets on data and monitoring and stresses the â€oeneed to take urgent
availability of disaggregated data to ensure that no one is left behindâ€. 26 The role of ICTS as a key development enabler
collect ICT data at the national level Monitoring progress towards achievement of the WSIS outcomes has been an integral component
indicators and collecting data and statistics on ICT. Since its creation in 2004, the Partnership has
national-level data on the indicators identified to measure the WSIS targets. The results of the survey are featured in the report
along with other data sources The Final review of the WSIS targets: Achievements, challenges and the way forward report is available at:
Using different data sources and big data analytics An important element in the discussion related
to the post-2015 development agenda and the setting of measurable goals and targets has
data and statistics are collected, analysed and disseminated, in view of the large data gaps prevailing in many developing countries in basic
statistics in the areas of the economy, health education, labour, etc. all of which are crucial to
see Box 1. 4 on the data revolution. Since then it has received considerable attention within the
development data According to the discussions, a data revolution considers new data sources, in addition to existing official sources:
besides governments other stakeholders such as the private sector, civil 27 Measuring the Information Society Report 2014
Box 1. 2: A decade of successful international cooperation on ICT measurement In 2004, ICT measurement was still in its infancy, with little
calls surfaced for reliable and comparable data in order to take stock of the emerging information society, identify digital
of their national ICT data collection. The methodological work developed by the Partnership has contributed significantly to
Data availability has increased also significantly over the past decade. In particular, there are more comparable data on ICT
infrastructure, household access and Internet users. For example at the beginning of the century, only around a dozen developing countries collected data on Internet users, while today there are
almost 50 developing countries collecting this indicator through official surveys (Chart Box 1. 2). Data on household access to
the Internet or a computer are now being collected by more than 100 economies worldwide, and data on Internet use in
businesses by almost 70 countries, although not on a regular basis (Partnership UNSC 2011. Similarly, whereas no data were
available on ICT access and use in schools, they have started now to be compiled in many developing countries (see section 1. 5
At the same time, major data gaps remain, in particular in developing countries and LDCS. This concerns, notably, statistics
public-sector organizations, ICT-related employment data, as well as data related to online security and cybercrime, gender
and youth, and cultural and environmental aspects. The growing information society will increasingly require more and better
Internet user data through official surveys, by level of development Note: Chart shows countries that have collected data on the number of
Internet users through official national surveys. Data are presented in three-year intervals and include countries that have collected
data for at least one year within those intervals Source: ITU 0 2 11 24 36
49 1 3 27 37 41 40 1 5 38 61 77 89 0 20
New data sources could include big data mostly provided by private-sector companies which could help â€oeimprove the timeliness and
completeness of data, without compromising the relevance, impartiality and methodological soundness of the statistics†(UNSC, 2014
The topic of big data is gaining momentum in the statistical community. Chief statisticians gathering at the UNSC meetings in 2013
recognized that â€oebig data constitute a source of information that cannot be ignored by official statisticians†and that â€oeofficial statisticians must
surveys in a number of countries, big data could provide important sources of more timely and
contributions from Member States. c Due to data limitations, currently mobile-broadband signal coverage is considering in determining this target
d Data being compiled by the Global Cybersecurity Index (GCI Source: ITU 29 Measuring the Information Society Report 2014
potentially becoming big data sources as well At the UNSC meeting in 2014, the commission reiterated its call for the global statistical
the use of big data for official statistics. The commission requested the group to include the
of big data for official statistics at regional subregional and national levels •To address the concerns of methodology
data and legislation related to big data •To address the issue of obtaining â€oeaccess at no cost†to big data from the private sector
for official statistical purposes, as well as the issue of access to transborder data or access to data on transboundary phenomena
•To develop guidelines to classify the various types of big data sources and approaches •To develop methodological guidelines
related to big data, including guidelines for all the legal aspects •To formulate an adequate communication strategy for data providers and users on the
issue of use of big data for official statistics •To reach out to other communities especially those more experienced
in IT issues or in the use of open data platforms The UN Global Working group on Big data for
Official Statistics was launched formally in June 2014, under the auspices of the UN Statistics Division.
The mandate of the group, of which ITU is a member, includes: provide strategic vision direction and coordination of a global programme
on big data for official statistics; promote practical use of sources of big data for official statistics
provide solutions for methodological, legal and privacy issues; promote capacity building; foster communication and advocacy of the use of big
data for policy applications; and build public trust in the use of private-sector big data for official
statistics ICTS are part of the debate on the data revolution big data and, more broadly, emerging data issues
in the post-2015 development debate. First the ICT sector in itself represents a new source
of data, provided by, for example, Internet and telecommunication companies. Second, the spread and use of ICTS allow public and private
entities across all economic sectors to produce store and analyse huge amounts of data. At the
same time, however, monitoring access to and use of ICTS by people, public entities and private
enterprises will be essential in order to identify the extent to which stakeholders in the ICT sector can be used as an alternative data source.
Without ICTS, no ICT-driven data revolution will take place In view of the link between big data and ICTS, work
is under way in ITU with a view to contributing to the debate and identifying new ways and means
of exploiting the potential of big data. The focus is primarily on the telecommunication/ICT sector
as a source of big data, including players such as operators and service providers, in the fixed
mobile and Internet sectors. Delegates attending the eleventh World Telecommunication/ICT Indicators Symposium (WTIS) in Mexico city in
December 2014 recommended that ITU should further examine the challenges and opportunities of big data, in particular data coming from ICT
companies; that regulatory authorities should explore the development of guidelines on how big data could be produced, exploited and stored
and that national statistical offices, in cooperation with other relevant agencies, should look into the opportunities for big data and address
current challenges in terms of big data quality Chapter 1. Recent information society developments 30 veracity and privacy within the framework of the
fundamental principles of official statistics. 33 The big data approach taken by ITU so far focuses on the following areas and questions
Standardization: 34 •Which standards are required to facilitate interoperability and allow technology integration in the big data value chain
•Which definitions, taxonomies, secure architectures and technology roadmaps need to be developed for big data analytics and technology infrastructures
•What is the relationship between cloud computing and big data in view of security frameworks •Which techniques are needed for data
anonymization for aggregated datasets such as mobile-phone records •How is exploited big data in different industries; what are the specific
challenges faced; and how can these challenges be addressed through international standards Regulation: 35 •What are the key regulatory issues at stake
and how can and should big data be regulated •How does big data impact on the
regulation of privacy, copyright and intellectual property rights (IPR transparency and digital security issues Box 1. 4:
What is a data revolution The report of the High-level Panel of Eminent Persons on the
Post-2015 Development Agenda to the UN Secretary-general which was published in May 2013, has called for, inter alia, a data
revolution taking advantage of new technology and improved connectivity: â€oewe also call for a data revolution for sustainable
development, with a new international initiative to improve the quality of statistics and information available to people and
levels on what such a data revolution could entail and how it could be implemented. While no internationally agreed concept
part of a data revolution: 32 •In view of the ubiquitous availability of communication networks, the use of new information technologies (e g
the collection and dissemination of data •Data should be further disaggregated (by gender income, age, geography, etc.
to ensure that no one is left out; in this regard, traditional statistical processes should be made more efficient
•The focus should go beyond data dissemination and also include investment in the development of concepts
•New, nontraditional data sources should be explored and leveraged to complement existing ones and satisfy the demand for data needs in new areas, such
as big data, geospatial information and geographical information systems •Open data policies should be envisaged to ensure
accountability and promote transparency •The role of data, statistics and monitoring for policy -making and decision-making should be increased
31 Measuring the Information Society Report 2014 •What is the link between big data and open data (crowdsourcing, cloud
computing, etc •is there need a to regulate data management and service providers •How can market dominance in the area
of big data be prevented and the rights of the data owners protected ICT data collection and analysis
•How can big data complement existing ICT statistics to better monitor information-society developments •Which type of data from ICT companies
are most useful and for which purposes •Which new ICT indicators could be produced from big data sources
•What are the key issues that need to be addressed, and by whom, in terms of
collecting and disseminating big data in telecommunications •What is the role of national statistical offices and how can big data
complement official ICT data •How can big data from telecommunications inform not only ICT but broader development policy in
real time, leading to prompt and more effective action Chapter 5 of this report addresses some of
these questions and provides suggestions and recommendations for the way forward Chapter 1. Recent information society developments
32 1 Refers to countries where fixed-telephone penetration increased by more than 1 per cent in 2014
2 See https://gsmaintelligence. com /3 http://www. censusindia. gov. in/2011census/hlo/Data sheet/India/Communication. pdf
4 4g refers to fourth-generation mobile network or service. It is a mobile-broadband standard offering both mobility and very
high bandwidth, such as long-term evolution (LTE) networks (ITU Trends 2014 5 Data collection on Europe and North america will follow in 2014
6 For a list of IXPS, see for instance http://www. datacentermap. com/ixps. html 7 For more details on international submarine fibre-optic links, see Telegeography†s Submarine cable Map 2014, available at
http://submarine-cable-map-2014. telegeography. com 8 For further discussion on progress made towards connecting rural households to the Internet,
IMF World Economic Outlook Database, April 2014 12 Source: The Economist, April 12 2014, â€oenigeria†s GDP step changeâ€
19 Data refer mostly to the year 2011 20 The UN E-government Development Index is a composite benchmarking indicator based on a direct assessment of the state
elements that should be part of a data revolution 33 See final report of WTIS-13, available at:
34 For further information on the work on big data carried out by the ITU Telecommunication Standardization Bureau (TSB),
http://www. itu. int/en/ITU-T/techwatch/Pages/big data-standards. aspx 35 A background document on big data that was prepared for GSR-14 is available at
http://www. itu. int/en/ITU-D/Conferences/GSR/Pages/gsr2014/default. aspx 35 Measuring the Information Society Report 2014
better data become available. For example what was considered basic infrastructure in the past †such as fixed-telephone lines †is fast
•Data availability and quality. Data are required for a large number of countries as the IDI is a global index.
There is relative paucity of ICT-related data, especially on ICT usage, in the majority of developing
countries. In particular, the three indicators included in the skills sub-index should be considered as proxies until
data directly relating to ICT skills become available for more countries •The results of various statistical analyses.
data and to explore whether the different dimensions are statistically well-balanced While the basic methodology has remained the
related data availability, the indicators included in the IDI and its sub-indices are under regular
field of ICT statistics and data collection, work through online discussion forums and annual face-to-face meetings.
-SIM ownership is driving up mobile-cellular subscription numbers, which is an issue in prepaid and, to a lesser extent, also in postpaid
data with data based on national household surveys (demand-side indicators. An indicator such as the percentage of individuals using a
reported these data to ITU for at least one year between 2011 and 2013. It is therefore too
subscription data in the IDI with mobile-phone user data. In view of the methodological difficulties in collecting harmonized data on
international Internet bandwidth, a review of the definition of the indicator is currently under discussion in EGTI
The IDI was computed using the same methodology as in the past, applying the following steps (Figure 2. 2 and Annex 1
•Preparation of the complete data set. This step includes filling in missing values using various statistical techniques
•Normalization of data. This is necessary in order to transform the values of the IDI indicators into the same unit of
•Rescaling of data. The data were rescaled on a scale from 0 to 10 in order to
compare the values of the indicators and the sub-indices •Weighting of indicators and sub-indices.
This chapter presents the IDI based on data from 2013 in comparison with 2012. It should be
•Country data revisions. As more accurate data become available, countries provide ITU with revised statistics for
previous years, which have been taken into consideration. This also allows ITU to identify inconsistencies and revise
added based on data availability. Overall this version of the IDI includes 166 countries/economies as compared with
According to data from the European union EU), 85 per cent of Danes have some level of
ITU World Telecommunication/ICT Indicators database 33 35 36 38 40 110 75 87 105
ITU World Telecommunication/ICT Indicators database have access to mobile broadband at speeds of at
Data also show that the Republic of korea achieves the highest advertised fixed-broadband speeds, with all
domestic demand for data driven by the high volume of local content, and domestic Internet
Data from the EU confirm that household access is highly correlated with regular use of the Internet, which
Data from the European Commission†s Digital Agenda underline the competitiveness of the European fixed
ITU World Telecommunication/ICT Indicators database 0. 0 0. 1 0. 3 1. 0 2. 5
ITU World Telecommunication/ICT Indicators database 91.5 88.1 97.2 96.4 0 10 20 30 40
mobile data plans. 35 During 2013, operators further extended their wireless infrastructure and services throughout Thailand,
Based on the 2013 and 2012 data presented in this chapter, the current (2013) global divide is
Data change very little over time and advances in skills do not show immediate effects
Data not available 59 Measuring the Information Society Report 2014 the profitability of various kinds of economic
selected for their high data availability for a large number of countries Past editions of this report (see MIS 2013) have
Data on urban population, population density, land area and GNI p. c. are sourced from the World bank.
Data on population size are sourced from UNPD.**Correlation is significant at 0. 01 level
pronounced when it comes to data on Internet access and use. Access to the Internet (be it
data are available for both sets of indicators. The following steps were performed •First, developed countries were excluded
in respect of which data pertaining to both the MDG indicators and the IDI are available for at least 16 countries. 41 The MDG
147 economies for which data were available for 2002 and 2013 shows that the global IDI value has doubled almost from 2. 52
would require different sets of data (including micro data) collected from official surveys Therefore, the analysis should be considered as a
first step in an attempt to quantify the relationship between ICT performance and MDG progress. The
data are available •Almost all MDG indicators that are included under MDG 1, MDG 4, MDG
children, through the analysis of health data collected through public health applications and by serving as a platform for exchange and
of smartphones to capture essential data on the patients and monitor their treatment has accelerated progress.
of MDG indicators, where 2002 and 2011 data are available for both sets of indicators.
data are available for 2002 and 2011 for both the MDG indicators and the IDI
earlier section where 2011 data were analysed improvements in the level of ICT access and use between the ten-year periods have shown
analysis will require different data sets, including micro data on ICT usage collected from official
surveys. Micro data offer analysts and researchers ample information and considerable flexibility to apply quantitative models that identify
relationships and interactions between indicators and topics covered in a survey, thereby fostering the diversity and quality of research and analysis
2 Data on the indicators included in the skills sub-index are sourced from the UNESCO Institute for Statistics (UIS.
http://data. worldbank. org/about/country-classifications/country-and-lending-groups#High income 7 Based on 2011 data:
https://ec. europa. eu/digital-agenda/sites/digital-agenda/files/DAE%20scoreboard%202013%20-%20
19 In these countries, the in-scope population for data on Internet users is aged individuals 16-74
as well as the household access data, excludes transient labourers, which account for a significant proportion of residents in Qatar.
According to data from ICTQATAR, â€oetransient labourers†make up 27 per cent of the overall population 33 http://qnbn. qa/qatar-vision-2030
35 http://www. telecompaper. com/news/thai-operators-reduce-prices-of-smartphone-data-plans--900198
World bank, see http://data. worldbank. org/indicator/NY. GNP. PCAP. CD 39 See: Czernich, N.,Falck, O.,Kretschmer, T. and Woessmann, L. 2009), Broadband Infrastructure and Economic growth
if there is available data for at least one year between 2010 and 2012. The number of countries included in the analysis varies from 16 to 101 developing countries.
since it is the closest year to available MDG data 42 Broadband Commission. Transformative Solutions for 2015 and Beyond a Report of the Broadband Commission Task force
ITU World Telecommunication/ICT Indicators database 0 20 40 60 80 100 120 140 160
Data from South African operators show that not only is wireless -broadband penetration reaching higher levels
consuming more data, indicating an increase in the intensity of usage. MTN reported a growth
of 63 per cent in data volumes in the first half of 2013 and Vodacom reported that on average
users were generating 75 per cent more data traffic per device than a year ago. 5 Wireless
and high levels of multi-SIM ownership (GSMA and Deloitte, 2013. Furthermore, the very high
Data from household surveys show that the actual number of people using a mobile-cellular phone is
ITU World Telecommunication/ICT Indicators database 2012 2013 P e r 1 0 0 i
Data from the annual ICT household survey show that, since 2008, computers have replaced telephones as
ITU World Telecommunication/ICT Indicators database 2012 2013 P e r 1 0 0 i
with typically high rates of multi-SIM ownership Furthermore, markets are very competitive, with a relatively high number of mobile operators.
Data from household surveys collected in a number of CIS countries underline that mobile -cellular penetration, measured as the number
ITU World Telecommunication/ICT Indicators database 2012 2013 P e r 1 0 0 i
Data from the Eurobarometer underlines this finding on average, 92 per cent of European union citizens (the majority of countries in the region
Data on Individuals using the Internet for Eurostat members are sourced from Eurostat. Eurostat collects data for Internet users aged 16-74 years old
Source: ITU World Telecommunication/ICT Indicators database World Developed 0 10 20 30 40 50
60 70 80 90 100 %101 Measuring the Information Society Report 2014 penetration stands at 6 per cent, all European
ITU World Telecommunication/ICT Indicators database 0 10 20 30 40 50 60 70 80
coverage, and the majority of all data traffic is carried by the LTE network. 27 Very high increases
11 Internet user data from Gulf countries are not comparable, as they refer to different populations.
Data from Bahrain and Qatar refer to the overall population, i e. including expatriate/transient workers.
Data from United arab emirates are estimated by ITU based on base data excluding the transient worker population 12 Reported in activated external capacity
13 http://www. ofca. gov. hk/en/industry focus/telecommunications/facility based/infrastructures/submarine cables/index html 14 http://submarinenetworks. com/systems/intra-asia/sjc/sjc-cable-system
/26 Data reported by the country refer to 2012 27 http://www. verizonwireless. com/wcms/consumer/4g lte. html and http://www. telecompaper. com/news/verizon-wireless
data following a harmonized methodology since 2008. Initially, prices were collected for fixed -telephone, mobile-cellular (voice and SMS) and
Since 2012, the data collection has been extended to include mobile -broadband prices. These data have proved to be
useful for the international comparison of ICT prices across more than 160 countries, and for identifying those cases where prices constitute a
They include end-2013 data for each of the three price sets contained in the IPB (fixed-telephone
Based on 140 economies for which 2008-2013 data on fixed-telephone and mobile-cellular prices were available
-broadband service with a 500 MB monthly data allowance. 7 Despite the limitations of comparing
included in the ITU price data collection exercise Growth rates have been stagnating since 2008 and there have been few structural changes in
price data had already fully or partially liberalized their fixed-telephone market in 2008, compared
Based on 140 economies for which 2008-2013 data on fixed-telephone and mobile-cellular prices were available
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Rank Economy Fixed-telephone sub-basket GNI p. c
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Rank Economy Mobile-cellular sub-basket GNI p. c
file sharing (less than 1 per cent of total file -sharing traffic was transmitted through mobile networks in 2013) and Internet video (2 per cent
total mobile data traffic was offloaded onto fixed networks in 2013 (CISCO, 2014), highlighting the role that fixed broadband plays in supporting the
which 2008-2013 data on fixed-broadband prices were available Source: ITU 0 50 100
and data on fixed (wired)- broadband subscriptions by speed, which show that a significant share
Based on 143 economies for which 2008-2013 data on fixed-broadband prices were available Source:
Based on 165 economies for which 2013 data on fixed-broadband prices were available Source: ITU
GNI p. c. values are based on World bank data 0. 0 0. 5 1. 0 1. 5
GNI p. c. values are based on World bank data 0 2 4 6 8 10 12
GNI p. c. values are based on World bank data for fixed-broadband uptake in the country.
GNI p. c. values are based on World bank data 0 5 10 15 20 A s
GNI p. c. values are based on World bank data There are 13 countries in the Asia and the Pacific
Indeed, the latest data on international connectivity show that this may remain an issue in Kiribati (45 Mbit/s), Marshall
GNI p. c. values are based on World bank data However, a comparison with other regions shows
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Rank Economy Fixed-broadband sub-basket
which ITU collects data on prices (Figure 4. 1 operators are adding plans for specific devices
allowing customers to pool the data consumed by different devices in a single subscription. 23
with neither time nor data volume constraints This is the common scheme in a vast majority
include data volume caps, e g. USD 10 for 50 MB per month. Several operators also offer
of time and data volume limitations, e g. USD 5 for one day of use with a maximum of 50 MB
priced on the basis of the data allowance (i e. the data volume in MB included in the plan) and not
the speed. Many operators do not even advertise the speed of the mobile-broadband service, but
in data allowances, bundled voice minutes and SMS, time limitations, premium speeds, etc rather than actual differences in prices for the
data allowance was PPP$ 36.6 (or USD 24.4) for prepaid plans and PPP$ 30.0 (or USD 19.2) for
-broadband service with 500 MB monthly data allowance was PPP$ 25.3 (or USD 16.9) for
the monthly data allowance was half as large Nevertheless, the reduction in price was not
proportional to the reduction in the data allowance, confirming that the price per GB is
lower for larger data allowances, the equivalent of a volume discount. Unlike in the case of
Based on 119 economies for which data on mobile -broadband prices were available for the four types of plans
data allowance of 500 MB are about eight times more affordable in developed countries than in developing countries, on average (Chart
service with 1 GB monthly data allowance corresponded to more than 20 per cent of GNI
Based on 119 economies for which data on mobile -broadband prices were available for the four types of plans
Based on 119 economies for which data on mobile -broadband prices were available for the four types of plans
much less than 500 MB of Internet data per month, supported by the fact that several African
limited data allowances, and even Internet radio would need to be limited This suggests that, if mobile broadband is to
Based on 119 economies for which data on mobile -broadband prices were available for the four types of plans
Country data for The americas reveal that there are a number of countries which have high prepaid computer
include a monthly data allowance much larger than 1 GB (Table 4. 8). This is the case of Chile
Such high monthly data allowances for prepaid mobile-broadband dongles suggest that these services target high-end customers
data allowances in The americas (Table 3. 7 suggesting that postpaid rather than prepaid is the base offer for regular computer-based
-broadband plans with a monthly data allowance of 1 GB suggest that mobile broadband could be
Percentages are calculated on the basis of the total number of countries with data available in each region:
data allowance of at least 1 GB. Although the minimum data allowance is the same, in practice
most fixed-broadband plans allow unlimited data use (Table 4. 4), whereas most computer -based mobile-broadband plans with a minimum
monthly data allowance of 1 GB really do have a cap of 1 GB (Tables 4. 7 and 4. 8
In almost half of the African countries included in the price benchmark, mobile-broadband prices were more than USD 10 cheaper per
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 1 Austria 0. 13 5. 31 4. 62 48'590 1'024
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 80 Maldives 2. 08 9. 7 12.6 5'600 700
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Chapter 4. ICT prices and the role of competition
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 1 Norway 0. 1 8. 34 5 102'610 500
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Rank Economy Mobile-broadband prepaid handset-based (500 MB
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 80 Antigua & Barbuda 2. 58 27.78 33.8 12'910 1'024
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 1 Austria 0. 13 5. 31 4. 62 48†590 1
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Rank Economy Mobile-broadband postpaid computer-based (1 GB
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 76 Montenegro 2. 89 17.49 28.93 7†260 3
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 1 Austria 0. 13 5. 31 4. 62 48†590 1
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates.**
ranked because data on GNI p. c. are not available for the last five years
GNI p. c. and PPP$ values are based on World bank data Rank Economy Mobile-broadband prepaid computer-based (1 GB
*Monthly data allowance (MB as%of GNI p. c. USD PPP$ 76 South africa 4. 82 28.90 51.17 7†190 1
to making available more data on household and individual economic welfare, as well as its distribution. This section presents a refined
data on the distribution of household income or consumption expenditure are available The objective is to explore how factors such as
Data on household income, on the other hand measure only people†s economic welfare, and do not include the business sector.
Data are collected by national statistical offices by means of household income and expenditure surveys HIES) or household surveys including a module
classify income data by deciles, individuals are placed in ascending order according to the household income attributed to them, and then
availability of products, may be relevant. 34 Data on household consumption expenditure are also obtained from household surveys
international comparisons based on data on household economic welfare, because there are several methodological issues that limit
-level economic data provide information (not available from macroeconomic indicators) on the actual income and expenditure capacity
previously mentioned, these data can be used to obtain a finer-grain indication of the affordability
Data for the United states and Sweden are sourced from the OECD Database on Income Distribution
Data for South africa and Viet Nam are sourced from the World Bank†s Povcalnet and refer to 2008
4. 18 uses data on income inequality to reveal differences in the affordability of fixed-broadband
Household disposable income based on World Bank†s Povcalnet data adjusted with ITU estimates on average persons per household
inequality within countries, economic data at the household level also make it possible to determine more precisely the affordability
Household consumption based on World Bank†s Povcalnet data adjusted with ITU estimates on average persons per household
and expenditure data, the following conclusions can be drawn •In more than 85 per cent of countries
for which data are available, the richest 20 per cent of the population can afford
data are available, a basic fixed -broadband subscription still represents more than 5 per cent of household
countries for which data on household income or expenditure distribution are available, fixed-broadband plans remain
Data on household disposable income refer to 2011 or latest year available.**†Lowest 20%†refers to the price divided by the average income of the first
Household disposable income for OECD countries and the Russian Federation based on data from the OECD Database on Income Distribution
on World Bank†s Povcalnet data adjusted with ITU estimates on average persons per household Fixed-broadband prices as a%of household disposable income%households
Data on household consumption expenditure refer to 2011 or latest year available.**†Lowest 20%†refers to the price divided by the average expenditure
Household consumption expenditure based on World Bank†s Povcalnet data adjusted with ITU estimates on average persons per household
limited by lack of data on the distribution of household income/expenditure Available data suggest that basic fixed
-broadband plans are affordable for 90 per cent of the population in Jordan and Tunisia, whereas they are less affordable in
with a 500 MB monthly data allowance) as a percentage of disposable household income and
collects price data (Chart 4. 13), and is currently the mobile-broadband service that is available in most
which data are available. This is also the situation in Armenia, Dominican republic and Egypt In these countries, a prepaid handset-based
subscription data show that, in developed countries, handset-based mobile-broadband subscriptions are individual, rather than shared
Data on household disposable income refer to 2011 or latest year available.**†Lowest 20%†refers to the price divided by the average income of the first
Household disposable income for OECD countries and the Russian Federation based on data from the OECD Database on Income Distribution
data adjusted with ITU estimates on average persons per household Prepaid handset-based mobile-broadband prices
Data on household consumption expenditure refer to 2011 or latest year available.**†Lowest 20%†refers to the price divided by the average expenditure
Household consumption expenditure based on World Bank†s Povcalnet data adjusted with ITU estimates on average persons per household
the household has his/her own SIM CARD with a mobile-broadband plan. In this case, affordability
Data show that handset-based mobile -broadband prices (with a 500 MB monthly allowance) are affordable for a majority of the
Data on equivalized household income and consumption expenditure per decile refer 2011 or latest available year
Equivalized household disposable income for OECD countries and the Russian Federation based on data from the OECD Database on Income
Equivalized household disposable income and consumption for other countries based on World Bank†s Povcalnet data adjusted with ITU
Data for the Arab States are limited to four countries. In Tunisia and Jordan, individual handset-based mobile-broadband subscriptions
On the basis of the data presented, it can be concluded that income inequality does not only
of comprehensive data series on the prices for these two services, which makes it possible to study
data for developing countries. Quantitative studies are in several cases restricted to samples of EU and OECD countries, or else
regulatory data for up to 144 countries in the period 2008-2013. Including such a large set
on telecommunication data from EU, OECD and specific countries hold true in a global context
the level of detail of the indicators, since data availability and comparability for such a large
Simple averages for 140 economies with available data on fixed-broadband prices and competition for the period 2008-2013
Herfindahl-Hirschman Index (HHI) data sourced from Informa Chart 4. 22: Competition in mobile markets and mobile
Simple averages for 140 economies with available data on mobile-cellular prices and competition for the period 2008-2013
Herfindahl-Hirschman Index (HHI) data sourced from Informa competition and the openness of the market to private and foreign investment
based on panel data regression. This enables us to go beyond descriptive statistics and draw some robust conclusions on the link between
analyses in this section, data from the Regulatory Tracker have been extracted for clusters 1, 2 and 3, and combined into a single value per country
for up to 144 countries based on data for the five-year period from 2008 to 2013
Panel data regression is a statistical technique which is used to assess how variations in a set
GB of data allowance included per month However, entry-level fixed-broadband plans in several countries offer higher
speeds and larger data allowances. In order to measure how these enhanced features affect prices, a variable on fixed
capped data allowances are included as controls in the model for fixed-broadband prices Chapter 4. ICT prices and the role of competition
Data collected by ITU, see Annex 2 for more details on the methodology for the collection of
Data collected by ITU, see Annex 2 for more details on the methodology for the
Data sourced from Informa Herfindahl-Hirschman Index for mobile cellular (voice and SMS with Li being the number of mobile-cellular subscriptions of firm i,
Data sourced from Informa 159 Measuring the Information Society Report 2014 Box 4. 2: Panel regression models for fixed-broadband and mobile-cellular prices (continued
Data collected by ITU, see www. itu. int/tracker for more information Fixed-broadband speed:
Speed of the entry-level fixed broadband plan in Mbit/s. Data collected by ITU
if the fixed-broadband plan includes a cap in the monthly data allowance, and 0 otherwise.
Data collected by ITU Descriptive statistics of the dependent variables Average Standard deviation Minimum Maximum 2008 2013 2008 2013 2008 2013 2008 2013
Descriptive statistics calculated for 124 economies that have complete data for the two models Source: ITU
Fixed-broadband plans with data caps are linked to prices 31 %lower than unlimited plans Fixed-broadband speed 0. 016
data caps by fixed-broadband service providers is correlated with cheaper entry-level fixed -broadband plans, other things being equal.
ITU data collection considers a minimum of 1 GB Chapter 4. ICT prices and the role of competition
data cap of 1 GB or more, or allows unlimited traffic. In 2013, entry-level fixed-broadband
plans included a data cap in 35 per cent of the countries included in the fixed-broadband
data caps were enforced in entry-level fixed -broadband plans. This suggests that operators can offer cheaper prices in exchange for reduced
data consumption, thus indicating that capacity in fixed-broadband networks is still an issue in several countries,
additional Internet data beyond 1 GB is still non -negligible in many countries Finally, different entry-level fixed-broadband
sector, such as operators†strategies on data caps competition in the fixed-broadband market and the ICT regulatory environment, may together be
is the existence of data caps, which is correlated with lower prices. This may indicate that fixed-broadband capacity
*Data correspond to the GNI per capita (Atlas method) in 2013 or latest available year adjusted with the international inflation rates
GNI p. c. and PPP$ values are based on World bank data Chapter 4. ICT prices and the role of competition
data caps, competition in the fixed -broadband market and the ICT regulatory environment, are together more of
information on the PPP methodology and data, see http://icp. worldbank. org 3 GNI takes into account all production in the domestic economy (i e.
http://data. worldbank. org/indicator/NY. GNP. PCAP. CD 4 Voice over internet services, such as Skype or Voipbuster, are excluded from the analysis in this section
9 Based on 2012 and 2011 ITU data for countries accounting for 97 per cent of global fixed (wired)- broadband subscriptions, it
http://data. worldbank. org/about/country-classifications/country-and-lending-groups 11 The Communications Commission of Kenya (CCK) issued mobile virtual network operator licences to three operators in April
These data suggest there is limited competition among operators 13 Advanced mobile technologies refer to standards agreed by the ITU Radiocommunication Assembly for next-generation
http://www. verizonwireless. com/wcms/consumer/shop/shop-data-plans/more-everything. html 24 The details of the different †4g†plans offered by Tigo can be found on the following websites
25 Data for mobile-broadband services have been collected since 2012 through the ITU ICT Price Basket Questionnaire,
http://data. worldbank. org/about/country-classifications/country-and-lending-groups 31 Household incomes include wages, salaries, self employment incomes, capital and property income, private transfers and
35 Differences in the equivalence scales of the source data used in this chapter are corrected roughly using ITU estimates on
World bank Povcalnet data on income distribution are published per capita, and figures from the OECD Database on Income Distribution are equivalized using the square root scale
36 For OECD countries, household income is estimated by multiplying the equivalized household income by the square
According to 2008/9 survey data, mean household consumption in Angola was almost the same as in
38 Data on income distribution are aggregated per household and then equally attributed to each member of the household
In addition, data on income distribution are averaged per decile. If the price of a fixed-broadband plan represents less
In conclusion, the data presented do not determine whether 100 per cent of a population within a country can afford a fixed-broadband plan,
The World Bank†s Povcalnet data are adjusted using ITU estimates on the average number of inhabitants per household
ITU Tariff Policies Database 2013 (ICTEYE http://www. itu. int/net4/ITU-D/icteye 48 For a detailed description of the ITU ICT Regulatory Tracker, see www. itu. int/tracker
market share data were available. This includes 95 economies from the developing world and 44 from the developed world
economies for which price and market share data were available for 2013. This includes 99 economies from the developing
share data were available. This includes 96 economies from the developing world and 44 from the developed world
Chapter 5. The role of big data for ICT monitoring and for development 5. 1 Introduction
-to date and reliable data, in particular from developing countries. The information and communication technologies (ICT) sector is
of those statistics and identify new data sources In this context, the emergence of big data holds
great promise, and there is an opportunity to explore their use in order to complement the existing, but often limited, ICT data
There is no unique definition of the relatively new phenomenon known as big data. At the most basic level it is understood as being data
sets whose volume, velocity or variety is very high compared to the kinds of datasets that
have traditionally been used. The emergence of big data is linked closely to advances in ICTS. In today†s hyper-connected digital world, people
and things leave digital footprints in many different forms and through ever-increasing data flows originating from, among other things
commercial transactions, private and public records that companies and governments collect and store about their clients and citizens, user
Big data have great potential to help produce new and insightful information, and there is a growing debate on how businesses
benefits of big data. Although it was the private sector that first used big data to enhance
efficiency and increase revenues, the practice has expanded to the global statistical community The United nations Statistical commission
NSOS) are looking into ways of using big data sources to complement official statistics and better meet their objectives for providing timely
value added by big data in the context of monitoring of the information society, and Chapter 5. The role of big data for ICT monitoring and for development
174 there is a need to explore its potential as a new data source. While existing data can provide
a relatively accurate picture of the spread of telecommunication networks and services there are significant data gaps
when it comes to understanding the development of the information society. Relatively little information for example, is available on the demand side
While an increasing number of countries currently collect data on the individual use of ICTS, many developing countries do not
produce such information (collected through household surveys or national population and housing censuses) on a regular basis Consequently, not enough data are available
about the types of activity that the Internet is used for, and little is known about the Internet
public services, even fewer data are available to show developments over time and enable informed policy decisions.
society, and that big data have the potential to help realize those efforts In addition to the data produced
and held by telecommunication operators, the broader ICT sector, which includes not just telecommunication companies but also over-the-top (OTT) service
captures a wide array of behavioural data Together, these data sources hold great promise for ICT monitoring,
and this chapter will explore the potential of today†s hyper-connected digital world to expand on existing access and
on the role and potential of big data when it comes to providing new insights for broader
Big data are already being leveraged to understand socio -economic well-being, forecast unemployment and analyse societal ties.
Big data from the ICT industry play a particularly important role because they are the only stream of big
data with global socioeconomic coverage. In particular, mobile telephone access is quasi -ubiquitous, and ITU estimates that by the end of
big data hold great promise for development However, while there are a growing number of research collaborations and promising proof
data for development, highlight advances, point to some best practices and identify challenges including in regard to the production and
sharing of big data for development The chapter will first (in Section 5. 2) describe some of the current big data trends and
definitions, highlight the technological developments that have facilitated the emergence of big data, and identify the main
sources and uses of big data, including the use of big data for development and ICT monitoring.
Section 5. 3 will examine the range and type of data that telecommunication companies, in particular mobile-cellular
operators, produce, and how those data are 175 Measuring the Information Society Report 2014 currently being used to track ICT developments
and improve their business. Section 5. 4 looks at the ways in which telecom big
data may be used to complement official ICT statistics and assist in the provision of new evidence for a host of policy domains
while Section 5. 5 discusses the challenges of leveraging big data for ICT monitoring and broader development, including in terms of
standardization and privacy. It will also make some recommendations for mainstreaming and fully exploiting telecom big data for monitoring
and for social and economic development in particular with regard to the different stakeholders involved in the area of big data
from the ICT industry 5. 2 Big data sources, trends and analytics With the origins of the term â€oebig data†being
shared between academic circles, industry and the media, the term itself is amorphous, with no single definition (Ward and Barker, 2013
Sources of big data Sources Some examples Administrative data •Electronic medical records •Insurance records •Tax records
Commercial transactions •Bank transactions (inter-bank as well as personal •Credit card transactions •Supermarket purchases
increasing amounts of data from different data sources. Indeed, one of the key trends fostering
the emergence of big data is the massive â€oedatafication†and digitization, including of human activity, into digital â€oebreadcrumbs†or
â€oefootprints†In an increasingly digitized world, big data are generated in digital form from a number of
sources. They include administrative records for example, bank or electronic medical records), commercial transactions between two
Big data is not just about the volume of the data. One of the earliest definitions, introduced
by the Gartner consultancy firm, describes big data characteristics such as velocity and variety, in addition to volume (Laney, 2001
â€oevelocity†refers to the speed at which data are generated, assessed and analysed, while the
Chapter 5. The role of big data for ICT monitoring and for development 176 term â€oevariety†encompasses the fact that data
can exist as different media (text, audio and video) and come in different formats (structured and unstructured.
The three-Vs definition has caught on and been expanded upon. A fourth V †veracity †was introduced to capture aspects
relating to data quality and provenance, and the uncertainty that may exist in their analysis (IBM
data (Jones, 2012)( Figure 5. 1 Included within the scope of big data is the category of transaction-generated data (TGD), 2
also sometimes described as â€oedata exhaust†or â€oetrace dataâ€. These are digital records or traces
that have been generated as by-products of doing things (such as processing payments making a phone call and so on) that leave behind
big data is that it is connected directly to human behaviour and its accuracy is generally high
Most of the data captured by telecommunication companies can be classified as TGD As is often the case with technological
The five Vs of big data Source: ITU at the forefront of extracting value from this
data deluge. Encouraged by promising results but also reduced budgets, the public sector is turning towards big data to improve its service
delivery and increase operational efficiency. In addition, there are uses for big data in broader development and monitoring, and there is an
increasing focus on big data†s role in producing timely (even real-time) information, as well as new insights that can be used to drive social and
Big data uses by the private and pub -lic sectors Marketing professionals, whose constant aim is to understand their customers,
preferences from the analysis of big data Walmart, the world†s biggest retailer, has been one of the largest and earliest users of big data
In 2004, it discovered that the snack food known as Pop Tarts was purchased heavily by United
of data generated through large-scale datafication and digitization of information Different types and forms of
data, including large amounts of unstructured data Level of quality accuracy and uncertainty of data and data
sources VALUE Potential of big data for socio -economic development VELOCITY Speed at which data are
generated and analyzed 177 Measuring the Information Society Report 2014 a specific condition that then led Walmart to
improve its production chain †in this case, by increasing the supply of Pop Tarts to areas likely
to be affected by a disaster. Walmart has also made use of predictive analytics, which uses personal information and purchasing patterns
to extrapolate to a likely future behaviour, and to better target and address customer needs
value of big data Nor is the private sector†s use of big data techniques restricted solely to market research
Companies and whole industries (healthcare energy and utilities, transport, etc. are using such techniques to optimize supply chains and
proficient in their use of data-driven decision making have been found to have productivity levels up to 6 per cent higher than firms making
minimal to no use of data for decision-making Brynjolfsson, Hitt and Kim, 2011. Significantly industries now have the ability to conduct
How big data saves energy †Vestas Wind Systems improves turbine performance Vestas, a global energy company dedicated to wind energy, with
installations in over 70 countries, has used big data platforms to improve the modelling of wind energy production and identify
By using big data techniques based on a large set of factors and an extended set of structured and
unstructured data, Vestas was able to significantly improve customer turbine location models and optimize turbine performance
Big data have enabled the creation of a new information environment and allowed the company to manage and analyse
weather and location data in ways that were previously not possible. These new insights have led to improved decisions
big data techniques to understand and control churn, optimize their management of customer relations and manage their network quality and
These fundamental shifts in data exploitation to generate new socioeconomic value coupled with the simultaneous emergence
of new rich data sources that can potentially be linked together and analysed with ease have sparked also the interest of governments
Encouraged by the potential of big data to produce new insights and slimmer budgets governments (at all levels) are now looking to
exploit big data and increase the application of data analytics to a range of activities, including
monitoring and improvement of tax compliance and revenues, crime detection and prediction and improvement of public service delivery
data they collect and generate themselves Chapter 5. The role of big data for ICT monitoring and for development
178 complement their official statistics by leveraging data from new sources, including crowd-sourced data generated by the public
In the United states, for example, Boston City Hall released the mobile app â€oestreet Bump†which uses a phone†s accelerometer to detect
potholes while the app user is driving around Boston and notifies City hall. 3 Some of the
richest data sources for enabling governments and development agencies to improve service delivery are actually external.
data include those captured and/or collected by the private sector, as well as the digital breadcrumbs left behind by citizens as they go
can make use of public and private databases and bigâ data analytics to improve public
administration, from land management to the administration of benefits. The Department of the Treasury has set up a â€oedo Not Pay†portal
which links various databases and identifies ineligible recipients to avoid wrong payments and reduce waste and fraud4 (The White house
Big data for development and ICT monitoring One of the richest sources of big data is the data captured by the use of ICTS.
This broadly includes data captured directly by telecommunication operators as well as by Internet companies and by content
providers such as Google, Facebook, Twitter etc. Big data from the ICT services industry are already helping to produce large-scale
development insights of relevance to public policy. Collectively, they can provide rich and potentially real-time insights to a host
of policy domains. It should be noted that in some countries and regions the use of big
data, including big data from the ICT industry is subject to national regulation. In the EU, for
example, a number of directives require data producers to obtain users†consent before gathering any of their personal data. 5
data in action and of the great potential of big data for broader development and monitoring
Mayer-Schã nberger and Cukier, 2013; Mcafee and Brynjolfsson, 2012. GFT worked by monitoring health-seeking behaviour expressed
focusing on the use of search-engine data to understand dengue fever outbreaks, 6 monitor prescription drug use (Simmering, Polgreen and
data beyond the realm of user search terms Online job-posting data are being used to
supplement traditional labour statistics in the United States7 and other countries. In another effort, an academic project at MIT known as the
data from hundreds of online retailers. 8 The data are used then by researchers to understand a
UN Global Pulse, a UN initiative to use big data for sustainable development and humanitarian action, has been mining Twitter data from
Indonesia (where Twitter usage is high) 9 to understand food price crises. Global Pulse was
analytics on the Twitter data to forecast the consumer price index several weeks in advance Byrne, 2013. As discussions on the post-2015
Pulse is also using Twitter data to understand and compare the relevance of different development topics among countries (Box 5. 2
Campaign are using big data and visual analytics to identify the most pressing development topics that people around the world
as a source of big data for monitoring purposes Regulators and others are now using the
Qos) data on broadband quality. For example the United states Federal Communications Commission (FCC) has released mobile apps that enable consumers to check their
Chapter 5. The role of big data for ICT monitoring and for development 180 Mobile data
Despite the rapid growth in Internet access 60 per cent of the world†s population is still
Depending on the source of Internet data results may also be biased more or less. A 2013 study into the characteristics and behaviour
is very similar to real life (census) data on the average age when American people get
-network big data seems to have the widest socioeconomic coverage in the near term and the greatest potential to produce
Mobile data are already being utilized for research and policy-making, not only in developed but also in developing economies
economic developments. 13 Data are also being used to improve responsiveness in the event of natural disasters or disease outbreaks.
data to understand the spread of malaria in Kenya (Pindolia et al. 2012; Wesolowski et al 2012a), and of cholera in Haiti after the 2010
Mobile network big data have been utilized to great effect in the area of transportation helping to measure and model people†s
data from the ICT sector, and especially those available to telecommunication operators have wide applicability for informing multiple
Leveraging such data to complement official statistics and facilitate broader development will enable governments as well as development agencies to better
big data with a view to understanding its potential for producing additional information and statistics on the information society
better understand the type of data that can be made available 5. 3 Telecommunication data
and their potential for big data analytics Fixed and mobile telecommunication network operators, including Internet service providers
ISPS), are an important source of data and for the purpose of this chapter, all forms of
telecommunication big data (either volume velocity or variety) are being considered. Most telecommunication data can be considered as
TGD, 14 that is, the result of an action undertaken such as making a call, sending an SMS,
accessing the Internet or recharging a prepaid card Since the service with the widest coverage and
-cellular service, data from mobile operators have the greatest potential to produce representative results and reveal developmental insights
Not surprisingly, the big data for development initiatives (outlined in Section 2. 2) have mainly drawn on mobile network big data rather than
on those from fixed-telephone operators or ISPS. Figure 5. 2 illustrates some of the similarities
Telecommunication data The mobile telecommunication data that operators possess can be classified into different types, depending on the nature of the information
they produce. They include traffic data, service access detail records, location and movement data, device characteristics, customer details and
tariff data. For a more detailed overview of these types of data, see Chapter 5 Annex
To collect traffic data, operators use a range of metrics to understand and manage the traffic
flowing through their networks, including the measurement of Internet data volumes, call, SMS and MMS volumes, and value-added service
VAS) volumes. Internet service providers can also use deep packet inspection (DPI), 15 which is a special process for scanning data packages
transiting the network Service access detail records, including call detail records (CDRS), are collected by operators
whenever clients use a service. They are used to manage the infrastructure and for billing purposes, and include information on the
time and duration of services used and the technology used, for example, for the mobile network (2g, 3g, etc..
These data are potentially also very useful for building a rich profile of customers, as outlined in this section
positioning data, with the latter providing more detailed and precise location information Since mobile user devices used to access mobile
Chapter 5. The role of big data for ICT monitoring and for development 182 Figure 5. 2:
An overview of telecom network data Source: ITU, adapted from Naef et al. 2014 Tr affi
Data volume Call volume SMS/MMS volume Erlang DPI data Timestamp of use Contact network
Duration of use Applicable charges Handset type Technology utilized (2g, 3g, DSL/ADSL, etc Billing address
Passive positioning data e g. cell ID IMEI Active positioning data e g. cell triangulation, GPS MAC address
Customer demographics (e g. age, gender, national ID card number Billing address Payment history (postpaid Billing address
Transaction generated data Stored warehouse data 183 Measuring the Information Society Report 2014 subscriber†s purchasing power (see below for
Finally, operators maintain tariff data in the form of billing records for their current and past
How mobile operators currently use data to track service uptake, business performance and revenues Operators use their TGD to monitor the uptake and penetration
On the basis of the detailed service-usage data collected telecommunication operators can produce a range of detailed
of originating and terminating calls, SMS and MMS usage, data upload volumes, data download volumes, level of use of different
VAS, and level of use of different OTT services. These data can be reported as averages (over time or for different categories of
user), as well as at various levels of aggregation (again over time or for different categories of user.
Finally, service consumption data are used to produce revenue data and projections at various levels of disaggregation or
aggregation. For example, the average revenue per user (ARPU) is a KPI for operators, which identify their most important customers
will, for instance, often associate revenue data with resource allocation to ensure that Qos at the base stations used by their
others, use aggregated revenue data to track and benchmark countries†ICT developments, monitor the evolution of the information society and
The telecom industry†s use of big data Telecommunication companies are actively seeking to intensify their use of big data analytics
in order to improve existing services and create new ones. For operators, big data open up opportunities for better understanding of their
customers, which in turn leads to improved sales and marketing opportunities. At the same time big data can help optimize network operations
and create new revenue streams and business lines, for example when selling data Customer profiling Telecom operators capture a range of
behavioural data about their customers Chapter 5. The role of big data for ICT monitoring and for development
184 Customer profiles include details about customers†mobility patterns, social networks and consumption preferences. Collectively, these
digital breadcrumbs enable operators to profile and segment their customers based on a variety of metrics (Figure 5. 3). Depending on the country
data regulations governing the manner in which operators may keep and/or use such data.
This being the case, the extent to which behavioural profiling is used by operators may vary greatly
Big data, on the other hand can help to enhance that classification by enabling analysis of the levels of
Big data techniques can help operators understand churn better by enabling them to model the likelihood
Customer profiling using telecom big data Source: ITU CUSTOMER INTERESTS SOCIO -ECONOMIC CLASS LEVEL OF INFLUENCE OF
subsidiary of SK TELECOM, uses big data to help its parent company to cut churn and generate new
revenue, and has used data mining to achieve a fourfold improvement in churn forecasting. The operator found that customers planning to quit
also seek to monetize the data they hold. The simplest way of doing this is to sell (anonymized
data to third parties. The customer insights obtained through the analysis of usage data can also help create new business lines, either
through innovation (e g. new types of VAS) or by partnering with other businesses, including credit-scoring and related financial services
One example is based the US big data startup Cignifi, 19 which obtains data from mobile operators and financial institutions to build credit
profiles and evaluate customer creditworthiness see Box 5. 8). Cross-promotions with brick-and -mortar businesses are a potentially high-growth
5. 4 Big data from mobile telecommunications for development and for better monitoring In 2013, the United nations High-level Panel of
on existing and new sources of data for the post-2015 development agenda (United nations 2013).
a report on â€oebig data and modernization of Chapter 5. The role of big data for ICT monitoring and for development
186 statistical systemsâ€, and proposed the creation of a big data working group at the global level
UNSC, 2013). 20 Current uses of big data to complement official statistics are still exploratory but there is a growing interest in this topic, as
evidenced by the numerous initiatives being pursued by the United nations, as well as by others, including the World bank, OECD, Paris21
There are many big data sources that can be used to monitor and assess development results. In a world where mobile telephony is
mobile telecommunication big data have unique potential as a new data source, with high mobile -cellular penetration levels and the increasing
use of mobile phones, even among the poorest and most deprived, making them particularly valuable by comparison with other types of
telecommunication data. Indeed, when referring to the data revolution, the United nations High-level Panel cited the example of â€oemobile
technology and other advances to enable real -time monitoring of development results†This section will present some of the existing
data in achieving development goals in various policy areas, including disaster management and sustainable and economic development
telecommunication big data have potential as a source to enable monitoring of the information society, although they have yet to assume a critical
potential of big data to complement its existing and often limited, set of ICT statistics. This section
big data could complement existing ICT indicators to provide a more complete comprehensive and up-to-date picture of the
Mobile phone big data for develop -ment Mobile data offer a view of an individual†s behaviour in a low-cost, high-resolution, real
-time manner. Each time a user interacts with a mobile operator, many details of the interaction
fact that these data are uniquely detailed and tractable, the information captured cannot easily be derived from other sources on such
The fact that the format of the data is relatively similar across different operators and countries creates a huge potential for
potential of mobile data for development in a number of different areas There have been a number of interesting
Big data for disaster management and syndromic surveillance21 Mobility data collected immediately after a disaster can in many cases help emergency
responders to locate affected populations and enable relief agencies to direct aid to the right
One application of such mobility data is for syndromic surveillance, especially to model the spread of vector-borne22 and
Using mobile data for development A recently published report (Cartesian, 2014) explores the potential of mobile data for development.
It points to three primary types of analysis-ex-post evaluation, real-time measurement, and future predictions and planning-in a
mobile data Migration monitoring Text analysis economic downturn prediction Text analysis commodity fluctuation prediction Assessment
Mobile data to track food assistance delivery Geo-targeted links between Ag suppliers /purchasers Pests, bad
positioning data with malaria prevalence data to identify the source and spread of infections Wesolowski et al.
showed how mobile phone data was used to track the spread of cholera after the 2010 earthquake (Bengtsson et al.
The integration of mobility data from mobile networks with geographic information frameworks, 23 supplemented with additional
Chapter 5. The role of big data for ICT monitoring and for development 188 Big data for better transportation planning
A data-centric approach to transportation management is already a reality in many developed economies.
Transportation systems are being fed with sensor data from a multitude of sources such as loop detectors, axle counters
parking occupancy monitors, CCTV, integrated public transport card readers and GPS data derived not only from phones but also from
public transport and private vehicles (Amini Bouillet, Calabrese, Gasparini and Verscheure 2011 One advantage of mobile networks is that
data, which, despite its limited spatial accuracy cell ID), has great potential for transportation planning.
data from mobile operator Orange to map out citizens†travel routes in Abidjan, the largest city in
and show how data-driven insights could be used to improve the planning and management of transportation services, thereby
Passive mobile positioning data has also been used for transportation planning and management in Estonia (Ahas and Mark, 2005
How mobile network data can track population displacements †an example from the 2010 Haiti earthquake
was produced on the basis of mobile network data to show the potential of big data in tracking population movements
Source: Bengtsson et al. 2011 Figure Box 5. 5: Tracking mobility through mobile phones Port-au-prince (Pap Number of
Leveraging mobile network data for transportation and urban planning in Sri lanka Very similar findings between the results of an official household
results of a big data analysis using mobile-phone data (left-hand map) underscore the merits of big data.
The image on the left based on mobile-phone data, depicts the relative population density in Colombo city and its surrounding regions at 1300
hours on a weekday in 2013, compared to midnight the previous day. While the yellow to red colouring shows areas in which the
Mobile big data (left) versus official survey data (right Both passive and active positioning data are used to
analyse traffic conditions, particularly in urban areas with higher base-station density. Active positioning data (especially GPS) produce higher precision in
location data and are therefore the most useful Operators may offer such specialized services based on passive or active location data) either
directly, or by providing data to third parties. Mobile network data are less expensive, are in real time and
are less time-consuming to produce than survey data, particularly in urban and peri-urban areas
where base-station density tends to be high In another example, the analysis of mobility
flows between two Spanish cities derived from three different data sources †mobile -phone data, geolocated Twitter messages and
the census †showed very similar results, and although the representativeness of the Twitter geolocated data was lower than the (real-time
mobile-phone and census data, the degrees of consistency between the population density profiles and mobility patterns detected by
means of the three datasets were significant Lenormand et al, 2014 Chapter 5. The role of big data for ICT monitoring and for development
190 Big data for socioeconomic analysis Data from mobile operators can provide insights in the areas of economic development and
socioeconomic status, often in near real time Big data techniques can therefore complement official statistics in the intervals between official
surveys, which are usually relatively expensive and time-consuming and therefore carried out infrequently. In many cases, insights derived
from big data sources may help to fill in the gaps rather than replace official surveys. It should
also be noted that mobile network big data are one of the few big data sources (and often the only one) in developing economies that
contain behavioural information on low-income population groups Frias-Martinez et al. 2012) developed a mathematical model to map human mobility
variables derived from mobile network data to people†s socioeconomic and income levels. The model took into account existing socioeconomic
and income-related data derived from official household surveys, and the results showed that populations with higher socioeconomic levels
are associated more strongly with larger mobility ranges than populations from lower socio -economic levels. By extending this method, the
study suggested that it was possible to create a model to estimate income levels based on data
from mobile network operators Another study, by Gutierrez, Krings and Blondel 2013), used two types of mobile network
data, namely subscriber communication data and airtime credit purchase records, to assess socioeconomic and income levels.
The authors used airtime purchase records based Box 5. 7: Poverty mapping in CÃ'te d†Ivoire using mobile network data
In CÃ'te d†Ivoire, researchers used mobile network data specifically communication patterns, but also airtime credit
purchase records) from Orange to estimate the relative income of individuals, as well as the diversity and inequality of income
levels. The research helped to understand socioeconomic segregation at a fine-grained level for CÃ'te d†Ivoire, with the
regarding revenue data and the difficulty this poses for outside parties wishing to obtain such
data The use of mobile-operator TGD can also foster financial inclusion by facilitating the provision of
example of how mobile big data can be used for the unbanked is Cignifi, a big data startup that uses
the mobile phone records of poor people to assess their creditworthiness when they apply for a loan
Big data for understanding societal structures Social-network studies relying on self-reporting relational data typically involve both a limited
number of people and limited number of time points (usually one. As a result, social-network
patterns based on telecommunication data it has become possible to obtain insights into societal structures on a scale that was
Using mobile-phone data to track the creditworthiness of the unbanked Cignifi, a big data startup, has developed an analytic platform
to provide credit and marketing scores for consumers, based on their mobile-phone data. The Cignifi business model is founded
on the idea that â€oemobile phone usage is not random †it is highly predictive of an individual consumer†s lifestyle and riskâ€
phone calls, text messages, data usage and, extrapolating from these, lifestyles †the company identifies patterns and uses
respond to changes in customer activity as the data are refreshed usually every two weeks. In addition to updating a person†s
findings from the model against historical lending data from approximately 40 000 borrowers using the mobile operator Oi†s
Chapter 5. The role of big data for ICT monitoring and for development 192 have been used to study the geographic
However, telecommunication data are also revolutionizing the study of societal structures at the micro level.
is possible to assess friendship using data from mobile network operators, and that the accuracy is compared high when with self-reported
data. Leveraging these behavioural signatures to obtain an accurate characterization of relationships in the absence of survey data could
also enable the quantification and prediction of macro and micro social-network structures that have thus far been unobservable
Big data to monitor the information society There is a case to be made for analysing data captured by telecommunication operators in
the interests of improving the current range of indicators used for monitoring the information society.
big data analytics. 25 The core indicators on ICT infrastructure and access include indicators on mobile-cellular and
mobile-broadband subscription data is that they do not refer to unique subscriptions, or mobile
to monitor the time during which a SIM CARD remains inactive. In June 2014, for example GSMA estimated that, globally, the number of
Survey-based data, for example on Internet users and mobile-phone users, do not entail the same issues as subscription data.
They are collected through household surveys, directly from citizens, and their level of reliability is relatively
Survey-based data can also be broken down by individual characteristics, including gender, age, educational level and occupation
these data is that they are not widely available in particular, many developing countries do not yet collect data on individual use of
or household access to, ICTS), are relatively expensive to produce, and are much less timely
than subscription data (often with a time lag of one year. Consequently, data on users of
the information society and the types of online service they consume are limited, and in many
networks and mobile big data could be used to identify alternative, less costly and faster ways of
administrative data from operators and survey data collected by NSOS, it is particularly interesting to assess some of the ways in
which big data can be used to overcome the shortcomings of existing key ICT indicators and to provide additional insights into ICT access
and use, user behaviour, activities and also the individual user. Big data could help in obtaining
more granular information in several areas, and big data techniques could be applied to existing data to produce new insights.
In particular operators†big data could produce information in the following areas Individual subscriber characteristics: Additional categorization across both time and space are
possible for subscription indicators, and big data could provide additional information on gender, socioeconomic status and user location
Information on gender or age, for example, could be derived from customer registration information notwithstanding a number of challenges and
privacy issues, as discussed later in Section 5. 5 The socioeconomic status of the person linked
data techniques applied to users†consumption information, as well as other data contained in customer registration information.
In addition, the analysis of customers†mobility patterns will often allow for an understanding of important locations
All subscription data could provide information as to location. In the case of fixed-telephone and fixed-broadband
to other (administrative) databases in order to Box 5. 9: Using mobile big data and mobile networks for implementing surveys
An important measurement for assessing the development of the information society is the extent to which households
surveys for collecting the corresponding data, and the declining response rates where traditional surveys are concerned (Groves
collection of survey data. This could include targeting a wide variety of respondents covering the full spectrum of appropriate
using big data analytics. In 2011, for example, UN Global Pulse partnered with Jana, 27 a mobile-technology company,
Chapter 5. The role of big data for ICT monitoring and for development 194 create new information.
Particularly rich possibilities exist where data from mobile-cellular and mobile-broadband subscriptions are concerned, since they are
voice, data and VAS †over time, and the intensity of use. Mobile operators are able to provide
In addition, mobile-operator data could be combined with customer information from popular online services, such as Facebook
the profiles developed using data from online services with customer profiles generated from analyses of mobile-operator data.
This would require telecommunication operators, OTT providers and other Internet content providers to work together and share information
By applying big data techniques to survey data and administrative data from operators, new insights could be derived, in particular, in respect
of the following Subscriptions versus subscribers: Big data techniques could help extrapolate the actual number of unique mobile subscribers or users
rather than just subscriptions, by comparing subscription numbers to user numbers derived from household surveys,
account usage patterns or data from popular Internet companies such as Google or Facebook By linking data collected from different sources
and combining subscription data and usage patterns, a correlation algorithm could be developed to reverse engineer approximate
values for these indicators, in order to estimate user numbers in between surveys, and possibly in real time. This could be pursued in a similar
variables using mobile-phone usage data, as described in greater detail at the beginning of this section. It is important to note here that
data methods only complement existing surveys rather than replacing them completely (see Section 5. 5 for a further discussion of this
In sum, relatively simple big data techniques can help analyse and provide complementary information on existing ICT data, and provide
new insights into the measurement of the information society. This includes information on the use of different services and applications
SIM usage and the fact that users will in many cases be using ICT services from more than
combining data from surveys with big data to build new correlation and predictive analytic techniques Finally, it should be noted that the methods that
and for other big data for development projects, big data analysis cannot replace survey data, which is needed to build and
test correlations and to validate big data results While the opportunities discussed above present what is analytically possible, data access
and privacy considerations are complex and nuanced, and therefore place constraints on what is practically feasible or advisable.
Section 5. 5 discusses these challenges in greater detail 5. 5 Challenges and the way
forward Attempting to extract value from an exponentially growing data deluge of varying structure and variety comes with its share of
challenges. The most pressing concerns are those associated with the standardization and interoperability of big data analytics, as
well as with privacy and security. Addressing such privacy and other concerns with respect to data sharing and use is critical,
and it is important for big data producers and users to collaborate closely in that regard. This includes
raising awareness about the importance and potential of producing new insights, and the establishment of public-private partnerships
to exploit fully the potential of big data for development Data curation, standardization and continuity Data curation and data preparation help to
structure, archive, document and preserve data in a framework that will facilitate human understanding and decision-making.
Traditional curation approaches do not scale with big data and require automation, especially since 85 per cent of big data are estimated to be
unstructured (Techamerica Foundation, 2012 Dealing with large heterogeneous data sets calls for algorithms that can understand the data
shape while also providing analysts with some understanding of what the curation is doing to
the data (Weber, Palmer and Chao, 2012 Telecom network operators themselves have to contend with interoperability issues arising
from the different systems (often from different vendors) they employ. It is not uncommon for
operators to write customized mediation software to overcome potential inter-comparability issues among data from different systems.
The problems are compounded when one has to take account of secondary third-party users that may seek to
leverage the data. The framework used by an NSO to organize data would be different from that
used by a network engineer or a marketing or business intelligence specialist. Naturally, telecom network operators have curated their data
based on their needs. To be able to use telecom big data for development and monitoring and to guarantee its continuity, the creation of
a semantic framework would require greater consensus among the many diverse stakeholders involved (telecom operators, network equipment
manufacturers, system developers, developmental practitioners and researchers, NSOS, etc Chapter 5. The role of big data for ICT monitoring and for development
196 Accessing and storing data, and data philanthropy Big data for development is still in its nascent
stages and, as such, comes with its share of challenges, not least of which is obtaining access to what is essentially private data.
Private corporations would hesitate to share information on their clients and their business processes in case such sharing is illegal, precipitates a
loss of user confidence and/or accidentally reveals competitive business processes. More importantly, companies will not share until there
Until holders of big data become more comfortable about their release it is going to be difficult for third-party research
big data, but it has taken them considerable time to build and leverage the necessary relationships with operators.
the necessary parameters as to how data are to be used, including the manner in which they
challenges associated with extraction of the data on account of different curation approaches and problems relating to the interoperability of
towards sharing data more publicly. Orange for example, hosted a â€oedata for Development Challenge, †releasing an aggregated anonymized
using Orange data from Senegal, is planned for 2015.30 In 2014, Telecom italia initiated a similar challenge, making data from the territories of
Milan and the Autonomous Province of Trento available to researchers for analysis. 31 It has
partnered with other data providers to curate and release additional big datasets containing weather, public and private transport, energy
event and social network data. In both the Orange and Telecom italia cases, researchers had to go through an approval process in order to
regular and safe sharing of data by building on the precedents being created by the ad hoc
the efforts being made to open up private-data stores in order to obtain actionable development insights There is a gap that needs to be addressed if
large-scale pooling and sharing of such data are to become a reality. Cross-sector and cross
and data-curation practices when pooling data from multiple sources. This facilitatory role may even be played by a third-party organization
faced by data providers, effectively acting as a gatekeeper to ensure that data are used transparently and in a way that contributes to
overall scientific knowledge generation, while ensuring that any safeguards that may be applicable in respect of private information are
trial data. To facilitate the process, they hired Yale University†s Open Data Access (YODA) Project
197 Measuring the Information Society Report 2014 to act as gatekeepers (Krumholz, 2014. YODA undertakes the necessary scientific review of any
make use of the data and ensures that necessary privacy and data usage guidelines are followed. 32
The question remains as to who is placed best to act as gatekeepers and standard-bearers when it comes to telecom network big data
Some have argued that NSOS are placed well to ensure that best practices are followed in the collection and representation of big data
and to provide a stamp of trust for potential third-party data seekers. Telecom operators for their part, are regulated mostly by sector
-specific regulators who can also have purview and dictate terms governing the privacy and data-reporting responsibilities of operators
Ultimately, however, the decision as to who takes on the gatekeeper and standardization function requires the confluence of multiple
to play in building an institutional model for data sharing and collaboration, in consultation with all
protocols) of privately held data such as mobile -phone records can be mutually beneficial to both government and private sector.
-action system shows how data sharing could be considered a business risk mitigation strategy for operators in emerging markets.
big data is linked closely to advances in the ICT sphere, including the falling cost of data storage
Depending on the data volume, storage can still be costly, especially where privacy considerations preempt the use of specialized third-party
cloud-based services. But as storage prices continue to fall, they are expected to be less of
As social scientists look towards private data sources, privacy and security concerns become paramount. To mitigate the potential risks, all
such data sharing. These stakeholders include not just the public and private sectors, but also significantly, the general public, who in many
cases are the primary producers of such data through their activities. It is also the public that
must ultimately decide on how the data they produce may be used. The World Economic Forum†s â€oerethinking Personal Data†project
new data ecosystem. 33 Discussions must address the individual†s privacy expectations, as well as those of private-sector stakeholders looking
identifiable individual (data subject) †(OECD 2013). ) The result of such an approach has been the policy of â€oeinform and consent†practised by
most companies to inform users of what data are Chapter 5. The role of big data for ICT monitoring and for development
198 being collected and how they will be used. It has been argued, however, that in a big data world
the â€oeinform and consent†approach is woefully inadequate and impractical, and that a new approach is needed (Mayer-Schã nberger and
data paradigm, the greatest potential often lies in secondary uses, which may well manifest long after the data was collected originally.
It is thus impractical for companies to have a priori knowledge of all the potential uses and to seek
Given the volumes of data that individuals are now generating, companies would find themselves struggling to maintain
privacy issues that may arise when data from one source is combined with data from other sources
to reveal/infer new data and insights. This blurs the lines between personal and nonpersonal
information, allowing seemingly nonpersonal data to be linked to an actual individual (Ohm 2010). ) Digitized behavioural data crumbs may in
fact greatly diminish personal privacy. The use of DPI, for example can technically reveal all of a user†s online activity.
Going one step further it is possible to understand a person†s needs behaviours and preferences by using data
-mining techniques on the digital breadcrumbs For instance, a recent study showed how Facebook â€oelikes†could accurately predict a range
Data anonymization35 (i e. methods designed to strip data of personal information), employed by computational social scientists, has been
called into question (Narayanan and Shmatikov 2008). ) A recent study of mobile CDRS for 1. 5
out that the data could in fact be de-anonymized completely by cross-referencing them with other
data sources. The attendant privacy concerns about such cross-referencing are clear, and have to be taken seriously and addressed
on the country, SIM resellers may pre-register the SIMS they sell under their own name, and
SIMS that are registered by one family member may be used by other members of the same family.
Sri lankan operators, for example, see a great mismatch between the person registering a subscription and the person using it.
data are kept secure. Data breaches undermine consumer confidence and hinder efforts to exploit big data for the greater social good.
Encryption virtual private networks (VPNS), firewalls, threat monitoring and auditing are some potential technical solutions that are employed currently
199 Measuring the Information Society Report 2014 but they need to be mainstreamed (Adolph 2013).
of big data, the World Economic Forum (WEF initiated a global multi-stakeholder dialogue on personal data that advocated a principle-based
approach that shifts governance from the data per se to its use; acknowledges the importance of context rather that treating privacy as a binary
privacy and data protection in a big data world the danger is that these questions may take too
use of big data for broader development. Hence a balanced risk-based approach may be required in the context of what is under discussion here
i e. the use of telecom big data for monitoring and development. This does still require the
of big data for development can be â€oesandboxed†with appropriate privacy protections imposed on researchers, while still ensuring that the broader
input data. In the big data paradigm, it is easy to overlook that concept, given the expectation
that when dealing with vast volumes of (often unstructured) data from a multitude of sources
â€oemessiness†is to be expected. As Mayer -Schã nberger and Cukier (2013) note, â€oewhat we lose in accuracy at the micro level we gain
Data quality and their provenance do matter, and the question is important in establishing the
generalizability of the big data findings Data provenance and data cleaning Understanding data provenance involves tracing the pathways taken by data from the
originating source through all the processes that may have mutated, replicated or combined the data that feed into the big data analyses.
This is no simple feat. Nor, given the varied sources of data that are utilized, is it always as feasible as
the scientific community would wish. However at the very least it is important to understand
some aspects of the origin of data. For example the fact that some mobile network operators choose to include the complete routing of a call
that has been forwarded means that there may be multiple records in the CDRS for the same
call. If that is not taken into consideration, the subsequent social network analysis could contain errors (overstating or understating tie strength
establish data provenance as envisaged by scientists, it is at the very least important to
have created the data Data cleaning remains a key part of the process to ensure data quality.
It is important to verify that the quantitative and qualitative (i e categorical) variables have been recorded as
expected. In a subsequent step, outliers must be removed, using decision-tree algorithms or other techniques.
However, data cleaning itself is a subjective process (for example, one has to decide which variables to consider) and not a
Are the data representative Related to the question of data provenance is the issue of understanding the underlying
Chapter 5. The role of big data for ICT monitoring and for development 200 population whose behaviour has been captured
The large data sizes may make the sampling rate irrelevant, but they do not necessarily
make it representative. Not everyone uses Twitter, Facebook or Google. For example ITU estimates suggest that 40 per cent of the
considering how telecommunication data may be used for monitoring and development While the potential benefits to be gained from
leveraging mobile network operator data for monitoring and development purposes hinges on the large coverage, close to the actual
data does not automatically mean that issues such as measurement bias and methodology internal and external data validity and data
interdependencies can be ignored. These are fundamental issues not just for â€oesmall data†but also for â€oebig data†(Boyd and Crawford, 2012
personas, so studying people†s data exhaust may not always give us insights into real-world dynamics.
where in essence the data artefact is itself a by-product of another activity. Telecom network big data, which mostly fall under this category
may be less susceptible to self-censorship and persona development, but the possibility of these phenomena cannot be ruled out.
In a way, big data analyses of behavioural data are subject to a form of the Heisenberg uncertainty principle, whereby
as soon as the basic process of an analysis is known, there may be concerted efforts to exhibit different behaviour and/or actions to change
considering big data analyses for monitoring purposes. Dr Nathan Eagle, a pioneer in the use of cellphone records to understand phenomena
when CDR data from Rwanda showed low mobility in the wake of flooding he theorized that this was due to an outbreak
-data analyses based on big data. For example prior research had established a power-law distribution between the frequency of airtime
the big data paradigm, leading to the discovery of misleading patterns. As Google†s Chief Economist, Hal Varian, notes, â€oethere are often
) Big data draws many of its techniques from machine learning, which is primarily about correlation and predictions. 40 Big data are by
their very nature observational and can measure only correlation and not causality. Supporters of big data have predicted the end of theory and
hypothesis-testing, with correlation trumping causality as the most relevant method (Anderson 2008; Mayer-Schã nberger and Cukier, 2013
mining of big data sources) will not drown out traditional deductive science (i e. hypothesis testing), even in a big data paradigm.
Among the three Vs in the traditional big data definition volume and variety produce countervailing forces.
More volume makes big data induction techniques easier and more effective, while more variety makes them harder and less effective
It is this variety issue that will ensure the need for explaining behaviour (i e. deductive science
Chapter 5. The role of big data for ICT monitoring and for development 202 rather than merely predicting it (Mullainathan
Causal modelling is possible in a big data paradigm by conducting experiments. Telecom network operators themselves use such
with â€oesmall data, †in this case the statistics collected by the Centers for Disease Control and
note, when combined with small data, â€oegreater value can be obtained by combining GFT with other near†real time health data. †Where data
from mobile network operators are used for syndromic surveillance, as in the case of malaria in Kenya (Wesolowski et al.
2012a), big data are most useful as a basis for encouraging timely investigation, rather than as a replacement
how telecommunication network data could be used for monitoring, surveys and supplemental datasets will remain important to sharpen the
prior to data anonymization to build a training dataset. This enabled them, for example, to understand variations in mobility, social networks
Virseda (2012) needed census data to build their algorithms and provide training data for their
algorithms to reverse engineer approximate survey maps. Official statistics will thus continue to be important to building the big data models
and for periodic benchmarking so that the models can be tuned fine to reflect ground realities Transparency and replicability
and replicability problems with big data research The fact that the original private data may in many cases not be available to everyone
underscores the importance of opening up such private-data sources (in a manner that addresses
potential privacy concerns) or of peer reviews that can hone and improve the analyses. Instead consumers of such research have no option but
data calls for a combination of specialized skills in the areas of data mining, statistics and
domain expertise, as well as data preparation cleaning and visualization. NSOS may have deep statistical skills in house, but this is not
enough when it comes to working with large volumes of big data calling for computer science and decision-analysis skills that are not
emphasized in traditional statistical courses Mcafee and Brynjolfsson, 2012. NSOS recognize this shortcoming. In a recent global survey of
data sources (UNSC, 2013. Currently, there is a mismatch between the supply of and demand for talented individuals with the requisite
i e. data scientists. Mckinsey predicts that by 2018 the demand for data-savvy managers and analysts in the United states will
amount to 450 000, whereas the supply will fall far short of this, at only 160 000 (Manyika et al
to leverage big data for development will face competition from the private sector when seeking to attract the right talent.
big data to complement official statistics, have a shortage of advanced analytical skills by comparison with developed economies.
Current research suggests that new big data sources have great potential to complement official statistics and produce insightful
of big data will have to overcome a number of barriers. This includes the development of models which protect user privacy while still
for using big data to complement official ICT statistics. Although this report highlights some of the big data sources and techniques that
could be used, further research is needed to understand and confirm the usefulness of big data sources for monitoring the information
society As with other official statistics, it is paramount for big data producers and big data users
to collaborate and to initiate a dialogue to identify opportunities and understand needs and constraints.
Since many of the big data sources lie within the private sector, close cooperation between NSOS, on the one hand
talk to commercial vendors of big data analytics In addition, operators and Internet companies can benefit greatly from engagement with
to leverage big data for different purposes Such engagement will also broaden their understanding of the limitations and assist them
the applications of data use for development are concerned, operators also have an interest in maximizing the economic well-being of their
Chapter 5. The role of big data for ICT monitoring and for development 204 privacy framework, in consultation with other
share their data with those from other sources including from competitors), but this is something that is worth exploring.
big data sources has great potential to increase added value and produce new insights. There is scope for exploring established models for such
some of their customer data with credit bureaux Governments Governments have different opportunities and different roles to play in the exploitation of big
data for monitoring and development. They can use big data to identify areas where rapid intervention may be necessary,
to track progress and make sure their decisions are based evidence and to strengthen accountability. More and more governments are recognizing the importance of
big data and have set up communities of practice and working groups to study their use and potential impact (UNSC 2013
setting big data standards. To this end, national regulatory authorities (NRAS) and NSOS, in consultation with other national stakeholders
big data clearing houses that promote analytical best practices in relation to the use of big data for complementing official statistics and for
development. Those standards, which NSOS are in the best position to enforce, would also have
to encompass best practices in relation to data curation and metadata standards. To this end NSOS must also prioritize the upgrading of the
handle big data, while at the same time investing in the necessary computational infrastructure As the main regulatory interface to the telecom
big data may be leveraged for social good Regulators have a role to play in facilitating the introduction of legislation that addresses privacy
concerns while encouraging data sharing in a secure manner. The following recommendations were made in a recently published ITU draft
In return, data users would be permitted to reuse personal data for novel purposes where a privacy assessment indicates
While the use of big data can help better decision-making through probabilistic predictions, this information should not be used against citizens
can utilize big data predictions •Fostering big data competition and openness: Regulators could foster big data competition in increasingly
concentrated big data markets, including by ensuring that data holders allow others to access their data under fair and
reasonable terms 205 Measuring the Information Society Report 2014 International stakeholders International stakeholders †including UN
agencies and initiatives (such as ITU and UN Global Pulse), the Partnership on Measuring ICT for Development, ICT industry associations
and producers of big data (Google, Facebook etc.)) †have an important role globally. More work is needed to understand fully the potential
of big data and examine the challenges and opportunities related to big data in the ICT sector. To this end, the key international
stakeholders have to work together to facilitate the global discussion on the use of big data
UN Global Pulse, as one of the main UN initiatives exploring the use of big data,
can do much to inform and motivate the discussion on global best practices and the use of big data for
development Where using big data for monitoring the information society is concerned, new partnerships, including public-private
partnerships between data providers and the ICT statistical community, including ITU, could be formed to explore new opportunities and
address challenges, including in the area of international data comparability and standards As one of the main international bodies working
on issues related to the telecommunication sector, ITU could leverage its position to facilitate global discussion on the use of telecom big data
for monitoring the information society Together, ITU and UN Global Pulse could facilitate the work that needs to be done by
NRAS and NSOS, through awareness raising and engagement on privacy frameworks, data sharing, and analytical global best practices
ITU could help reduce the transaction costs associated with obtaining telecommunication big data, for example by facilitating the
standards-setting process. Standardized contracts for obtaining data access as well as standards on how the data are stored,
collated and curated can collectively reduce the overall transaction costs of accessing and leveraging telecommunication big data for social good
Academia, research institutes and develop -ment practitioners The research into how telecom data may be used to aid broader development is being done
mainly by academia, public and private research institutes and, to a lesser degree, development practitioners. This makes them important
stakeholders in defining the state of the art with respect to leveraging big data for development They, more than others, have been the first to
engage with telecommunication operators with a view to using their data for development They therefore understand the potential and
challenges from multiple perspectives. Their collective experiences will be valuable as big data for development becomes mainstreamed
207 Measuring the Information Society Report 2014 Chapter 5 Annex The mobile-telecommunication data that
operators possess can be classified into different types, depending on the nature of the information they produce.
They include traffic data, service access detail records, location and movement data, device characteristics, customer
details and tariff data Traffic data Operators use a range of metrics to understand and manage the traffic flowing through their
networks. These include •Data volume: both uplink and downlink volumes for Internet traffic can be captured at various levels of
disaggregation down to the individual subscriber, or even to the level of a base station (in the case of a mobile operator
or local switch (in PSTN networks. These can be analysed to understand subscriber demand for data at both an individual
level and at aggregate levels, and the understanding thus gained can be used for billing purposes and for network
management •Erlang: a dimensionless metric used by mobile network operators to understand the offered and utilized network load. 41
Erlang data are used to understand the load on a base station at any given time
the entire data stream to be captured and stored, owing to the storage requirements that would be needed and also to
The most common use of such data is for basic billing purposes, in addition to which they can
Chapter 5. The role of big data for ICT monitoring and for development 208 Location and movement data
Mobile networks can, depending on their sophistication, capture a range of movement and location variables, which can be broadly
active positioning data (see Annex Box 1 Annex Box 1: Active versus passive positioning data
Passive positioning transaction generated data (TGD) is automatically generated by the network and captured in the
operator†s logs for billing and network management purposes to understand network load and to keep track of the handset in
Active positioning data (which is of relevance only to mobile networks) is location and movement data that is captured in response to a specially initiated network
query to locate a handset using either network or handset-based positioning methods. GPS location data can also be considered
as active positioning data Active positioning data Active positioning data can be generated using either device
-centric or network-centric methods, as well as via satellite i e. GPS. The use of these methods has developed either in
response to national regulations requiring operators to capture higher-precision location data, and/or to a business case for
providing location-based services. The large-scale capture of such higher-resolution data is undertaken mainly by operators
in developed economies. Operators in developing economies use some of these methods, but often on a case-by-case basis
and not for their entire subscriber base. 43 However, this trend is currently changing, and an increasing number of regulators are
Passive positioning data Passive location data are contained in the subscriber†s registration information, which includes some form of mailing address
billing address in the case of postpaid. While fixed-telephone network operators have access only to static location data
mobile networks have much richer and dynamic location data CDRS, SMS detail records and Internet access records are the
main sources of passive positioning data for mobile operators and reside in their data warehouses. These records include the
ID of the antenna (cell ID), which in turn has a geolocation an azimuth (i e. antenna orientation information) and an
angular tilt. 45 It is also possible to obtain such data in real time through data-mediation services,
but these are not universally implemented Passive location data from the billing records are obviously
sparse and generated only when the phone is used and when the network knows which cell a particular handset is currently
connected to. However, many operators choose not to archive these data if they do not have a business case to justify the
additional storage costs. Where they are archived, such cell -handoff data provide a time-stamped sequence of cells that the
phone was attached to, and provides for a rich mobility profile as compared to the event-based billing records
Passive positioning data based on cell IDS is inexpensive when compared to active positioning data, but the tradeoff
lies in their lower precision, usually at the level of network cells. This lower-resolution location estimate can range from
such passive location data, at an aggregate level (temporal and/or spatial) these data remain very valuable
Source: ITU Device characteristics All mobile user devices used to access mobile telecommunication services come with an
The Aoa method uses data from base stations that have been augmented with arrays of smart antennas.
Such location data have high spatial resolution, but are costly for operators to implement Enhanced observed time
items of demographic data during the customer registration process. These can include the customer†s age, gender, billing address and
Tariff data Operators maintain the complete tariff sheet and billing records for their current and past services
Mobile operators can associate such data with traffic data to understand the revenue that is being generated by specific network elements
e g. base stations), not only retrospectively, but also, possibly, in real time Chapter 5. The role of big data for ICT monitoring and for development
210 1 The report of the UN Secretary-General†s High-level Panel of Eminent Persons on the post-2015 Development Agenda
calls for a data revolution that â€oewould draw on existing and new sources of data to fully integrate statistics into decision-making
and use of, data and ensure increased support for statistical systems. †This suggests that efforts to
and complement, official statistics have turned to the search for new data sources, including big data. In addition, the European Statistical System Committee (ESSC) in 2013 adopted the Scheveningen Memorandum
on  Big data and Official Statisticsâ, which acknowledges that Big data represents new opportunities and challenges for
Official Statistics, and which encourages the European Statistical System and its partners to effectively examine the potential
of Big data sources, see: http://epp. eurostat. ec. europa. eu/portal/page/portal/pgp ess/0 docs/estat/SCHEVENINGEN
MEMORANDUM%20final%20version. pdf 2 This term was discussed first in 1991, although the term then used was generated â€oetransaction information†(Mcmanus, 1990
on the protection of individuals with regard to the processing of personal data and on the free movement of such data
to which personal data cannot legitimately be processed without the consent of the data subject, except if necessary to
of the data subject, who should be given clear and comprehensive information as to the manner and purpose of such
http://policyreview. info/articles/analysis/big data-big-responsibilities 6 See https://www. google. org/denguetrends
/7 A good example of this is the Conference Board Help Wanted Online (HWOL) data series that measures the number of new
http://www. conference-board. org/data/helpwantedonline. cfm 8 See http://bpp. mit. edu /9 According to Peerreach. com, 20 per cent of Indonesia†s online population uses Twitter, the second highest ratio in the world
/11 ITU World Telecommunication/ICT Indicators database, 17th edition, 2014, available at http://www. itu. int/en/ITU-D/Statistics/Pages/publications/wtid. aspx
15 Deep packet inspection (DPI) is a process that utilizes specialized software to scan all of the data packets traversing a
and analysis of health data about a clinical syndrome that has a significant impact on public health, with the data in question being used to drive decisions about health policy and health education
22 A vector-borne disease is a disease that is transmitted through an agent (person, animal or microorganism
23 A geographic information framework is a representation framework that codes the components of geospatial data in a
standardized manner so as to facilitate analyses and data exchange 24 See http://www. itu. int/en/ITU-D/Statistics/Pages/intlcoop/partnership/default. aspx for more information regarding the
30 More information about the Data for Development (D4d) challenges using Orange data can be found at
32 More information about the Yale university Open Data (YODA) project can be found at http://medicine. yale. edu/core/projects/yodap/index. aspx
42 DPI is a process that utilizes specialized software to scan all of the data packets traversing a particular IP network.
Chapter 5. The role of big data for ICT monitoring and for development 212 45 Most network operators use multiple sectorized antennas on a single base station.
47 An international mobile subscriber identity (IMSI) number is a 15-digit number unique to the particular SIM in a subscriber†s
the data deluge makes the scientific method obsolete. Wired Magazine, 16 (7 Askitas, N. and Zimmermann, K. F. 2009), Google econometrics and unemployment forecasting.
Azman, A s.,Urquhart, E. A.,Zaitchik, B. and Lessler, J. 2013), Using Mobile phone Data to Supercharge Epidemic Models of
Data for Development: Net Mobi 2013 Bengtsson, L.,Lu, X.,Thorson, A.,Garfield, R. and von Schreeb, J. 2011), Improved response to disasters and outbreaks by
tracking population movements with mobile phone network data: a post-earthquake geospatial study in Haiti. PLOS
mobility and optimizing public transport using cellphone data. Data for Development: Net Mobi 2013 Blumenstock, J. E. 2011), Using mobile phone data to measure the ties between nations.
Proceedings of the 2011 iconference pp. 195†202. New york, USA: ACM Press. doi: 10.1145/1940761.1940788
Blumenstock, J. E. 2012), Inferring patterns of internal migration from mobile phone call records: evidence from Rwanda
Bollier, D. 2010), The Promise and Peril of Big data. WASHINGTON DC Bouckaert, J.,Van dijk, T. and Verboven, F. 2010), Access regulation, competition, and broadband penetration:
Boyd, D. and Crawford, K. 2012), Critical questions for big data. Information, Communication & Society, 15 (5), 662†679. doi:
How Does driven Data Decisionmaking Affect Firm Performance? Retrieved from: http://papers. ssrn. com/sol3/papers. cfm?
Byrne, C. 2013), How The UN€ s New Data Lab in Indonesia Uses Twitter to Preempt Disaster, Fast Company.
http://www. fastcolabs. com/3007178/open-company/how-uns-new-data-lab-indonesia-uses-twitter
Caceres, N.,Wideberg, J. P. and Benitez, F. G. 2007), Deriving origin†destination data from a mobile phone network.
Calabrese, F.,Di Lorenzo, G.,Liu, L. and Ratti, C. 2011), Estimating origin-destination flows using mobile phone location data
Global Mobile Data Traffic Forecast Update, 2013-2018, White paper, February 2014 Available at: http://www. cisco. com/c/en/us/solutions/collateral/service-provider/visual-networking-index-vni/white
David, T. 2013), Big data from Cheap Phones. Technology Review, 116 (3), 50†54 De Montjoye, Y.-A.,Hidalgo, C a, Verleysen, M. and Blondel, V. D. 2013), Unique in the Crowd:
D. 2009), Inferring friendship network structure by using mobile phone data. Proceedings of the National Academy of Sciences of the United states of america, 106 (36), 15274†8. doi:
using search engine query data. Nature, 457 (7232), 1012†4. doi: 10.1038/nature07634 Groves, R. M. 2011), A possible data future for the observational social sciences.
COSSA 30th Anniversary meeting Washington, D c Gruber, H. and Koutroumpis, P. 2013), Competition enhancing regulation and diffusion of innovation:
GSMA (2013b), The Mandatory Registration of Prepaid SIM CARD Users, White paper. Retrieved from: http://www. gsma
com/publicpolicy/wp-content/uploads/2013/11/GSMA WHITE-Paper mandatory-Registration-of-Prepaid-SIM -Users 32pgwebv3. pdf References
Harford, T. 2014), Big data: are we making a big mistake? Financial times. pp. 7†11. Retrieved 28 march 2014 from
The real-world use of big data. How innovative enterprises extract value from uncertain data. Retrieved from:
http://public. dhe. ibm. com/common/ssi/ecm/en/gbe03519usen/GBE03519USEN. PDF IBM (2013), The Four V†s of Big data.
Retrieved from http://www. ibmbigdatahub. com/sites/default/files/infographic file/4-Vs-of-big data. jpg
ICT Qatar (2014), Qatar†s ICT landscape 2014 households and individuals. Available at: http://www. ictqatar. qa/en/documents
ITU (2011), Handbook for the collection of administrative data on Telecommunications/ICT. Available at http://www. itu. int/en/ITU-D/Statistics/Pages/publications/handbook. aspx
ITU (2013), Big data: Big today, normal tomorrow. Retrieved from: http://www. itu. int/oth/T2301000022/en
ITU GSR discussion paper (2014), Big data-Opportunity or Threat. Retrieved from: http://www. itu. int/en/ITU-D/Conferences
Krumholz, H. M. 2014), Give the data to the People, New york times. Retrieved 2 february 2014 from
http://www. nytimes. com/2014/02/03/opinion/give-the-data-to-the-people. html? hp&rref=opinion& r=1
Laney, D. 2001), 3d data management: Controlling data volume, velocity and variety. Gartner, 6 Lazer, D.,Kennedy, R.,King, G. and Vespignani, A. 2014), Big data.
The parable of Google Flu: traps in big data analysis. Science New york, N y.),343 (6176), 1203†5. doi:
10.1126/science. 1248506 Lazer, D.,Pentland, A.,Adamic, L.,Aral, S.,Barabasi, A l.,Brewer, D.,Van alstyne, M. 2009), Computational social science
Science (New york, N y.),323 (5915), 721†3. doi: 10.1126/science. 1167742 Lenormand, M, Picornell, M.,Cantu-Ros, O.,Tugores, A.,Louail, T.,Herranz, R.,Barthelemy, M.,Frias-Martinez, E.,Ramasco, J
Lokanathan, S.,Silva, N. de, Kreindler, G.,Miyauchi, Y. and Dhananjaya, D. 2014), Using mobile network big data for informing
Manyika, J.,Chui, M.,Brown, B.,Bughin, J.,Dobbs, R.,Roxburgh, C. and Byers, A. H. 2011), Big data:
Data for Development: Net Mobi 2013 Mayer-Schã nberger, V. and Cukier, K. 2013), Big data:
A revolution that will transform how we live, work, and think. Houghton Mifflin Harcourt Mcafee, A. and Brynjolfsson, E. 2012), Big data:
the management revolution. Harvard Business Review, 90 (10), 60†6, 68,128 Retrieved from: http://www. ncbi. nlm. nih. gov/pubmed/23074865
What big data means for social science. Headcon †13 Part I. Retrieved from http://edge. org/panel/headcon-13-part-i
Naef, E.,Muelbert, P.,Raza, S.,Frederick, R.,Kendall, J. and Gupta, N. 2014), Using mobile data for development.
Pindolia, D. K.,Garcia, A j.,Wesolowski, A.,Smith, D. L.,Buckee, C. O.,Noor, A m.,Tatem, A j. 2012), Human movement data
2012), Demystifying Big data: A Practical Guide To Transforming The Business of Government Retrieved from: http://www. techamerica. org/Docs/filemanager. cfm?
The White house (2014), Big data: Seizing Opportunities, Preserving values. Executive Office of the President. Retrieved from
Using New Data to Understand Emerging Vulnerability in Real-time UN High-level Panel (2013), A new global Partnership:
UNSC (2013), Report of the Secretary general on Big data and modernization of statistical systems UNSC (2014), Big data and modernization of statistical systems, December 2013.
Available at http://unstats. un. org/unsd/statcom/doc14/2014-11-Bigdata-E. pdf Van damme, E. 2002), The European UMTS-auctions, European Economic Review, Volume 46, Issues 4-5 may 2002, pp. 846
Varian, H r. 2013a), Beyond Big data. NABE Annual Meeting. San francisco, CA Varian, H r. 2013b), Big data: New Tricks for Econometrics
Wallsten, S. 2001), An econometric analysis of telecom competition, privatization, and regulation in Africa and Latin america
A. 2013), Undefined By Data: A Survey of Big data Definitions, 2. Databases. Retrieved from http://arxiv. org/abs/1309.5821
Waverman, L. and Koutroumpis, P. 2011), Benchmarking telecoms regulation: The Telecommunications Regulatory Governance Index (TRGI.
Weber, N m.,Palmer, C. L. and Chao, T. C. 2012), Current Trends and Future Directions in Data Curation Research and Education
Rewards and Risks of Big data (B. Bilbao-Osorio, S. Dutta and B Lanvin, Eds References
Using Cellular network Data Data for Development: Net Mobi 2013 Xia, J. 2011), The third-generation-mobile (3g) policy and deployment in China:
Current status, challenges, and prospects Telecommunications Policy, Volume 35, Issue 1 february 2011, pp. 51-63. doi:
objectives, data availability and the results of various statistical analyses such as the principal component analysis (PCA.
Data for all of these indicators are collected by ITU. 2 1. Fixed-telephone subscriptions per 100
data cards or USB modems, subscriptions to public mobile data services, private trunked mobile radio, telepoint, radio paging and
telemetry services 3. International Internet bandwidth (bit/s) per Internet user International Internet bandwidth refers to the
bandwidth, in megabits per second (Mbit/s It is measured as the sum of used capacity of
Data are obtained by countries through national household surveys and are provided either directly to ITU by national statistical offices
There are certain data limits to this indicator insofar as estimates have to be calculated for many developing countries which do not yet
more data become available, the quality of the indicator will improve 5. Percentage of households with Internet
Data are obtained by countries through national household surveys and are provided either directly to ITU by national statistical offices
There are certain data limits to this indicator insofar as estimates have to be calculated for many developing countries which do not yet
more data become available, the quality of the indicator will improve b) ICT use indicators
Data for all of these indicators are collected by ITU. 4 1. Percentage of individuals using the Internet
Data are obtained by countries through national household surveys and are provided either directly to ITU by national statistical offices
There are certain data limits to this indicator insofar as estimates have to be calculated for many developing countries which do not yet
more data become available, the quality of the indicator will improve 2. Fixed (wired)- broadband subscriptions per
mobile-broadband data subscriptions to the public Internet. It covers actual subscribers, not potential subscribers
with advertised data speeds of 256 kbit/s or greater that allow access to the
been used to set up an Internet data connection using Internet Protocol (IP) in the past three months.
Internet data connection, even if the messages are delivered via IP. Dedicated mobile-broadband data subscriptions refers to subscriptions to dedicated data
services (over a mobile network) that allow access to the greater Internet and which are purchased separately from
voice services, either as a standalone service (e g. using a data card such as a USB modem/dongle) or as an add-on data
package to voice services which requires an additional subscription. All dedicated mobile-broadband subscriptions with
recurring subscription fees are included regardless of actual use. Prepaid mobile -broadband plans require use if there is no
Data on adult literacy rates and gross secondary and tertiary enrolment ratios are collected by the
data A critical step in the construction of the index is to create a complete data set, without missing values
There are several imputation techniques that can be applied to estimate missing data. 6 Each of the
imputation techniques, like any other method employed in the process, has its own strengths and
ensure that the imputed data will reflect a country†s actual level of ICT access, usage and skills
missing data, where previous year data are not available to calculate the growth rates. Hot -deck imputation uses data from countries with
â€oesimilar†characteristics, such as GNI per capita and geographic location. For example, missing data for country A were estimated for a certain indicator by
first identifying the countries that have similar levels of GNI per capita and that are from the same region
use data of country A was estimated by using Internet use data of country B from the same region
with similar level of GNI per capita and similar level of fixed Internet and wireless-broadband
estimate missing data for all indicators included in the index 3. Normalization of data Normalization of the data is necessary before any
aggregation can be made in order to ensure that the data set uses the same unit of measurement For the indicators selected for the construction
of the IDI, it is important to transform the values to the same unit of measurement, since
some values are expressed as a percentage of the population/of households, whereby the maximum value is 100, while other indicators
of the huge dispersion of values, the data were transformed first to a logarithmic log) scale. Outliers were identified then
After normalizing the data, the individual series were rescaled all to identical ranges, from 1 to
the data were transformed first to a logarithmic (log) scale The ideal value of 787†260 bit/s per Internet user is equivalent to 5. 90 if transformed to a log scale
weighting and aggregation of the data Each of the processes or combination of processes affects the IDI value.
1 Principal component analysis was used to examine the underlying nature of the data. A more detailed description of the
2 More information about the indicators is available in the ITU †Handbook for the collection of administrative data on
EGH. pdf). As some of the data used in the calculation of the IDI were collected before that meeting,
however, the data may not necessarily reflect these revisions 4 More information about the indicators is available in the ITU â€oehandbook for the collection of administrative data on
telecommunications/ICT€ †2011, see ITU 2011b and the ITU â€oemanual for Measuring ICT Access and Use by Households and
Annex 2. ICT price data methodology 1. Price data collection and sources The price data presented in this report were
collected in the fourth quarter of 2013. The data were collected through the ITU ICT Price Basket
questionnaire, which was sent to the administrations and statistical contacts of all 193 ITU Member States in October 2013.
Through the questionnaire contacts were requested to provide 2013 data for fixed-telephone, mobile-cellular, fixed-broadband
and mobile-broadband prices; the 2011 and 2012 prices were included for reference, where available. For those countries that did not reply
The collection of price data from ITU Member States and the methodology applied for the IPB was agreed upon by the ITU Expert
Annex 2. ICT price data methodology 232 The fixed-telephone sub-basket does not take
for a SIM CARD. The basket gives the price of a standard basket of mobile monthly usage in USD
6. The same price plan should be used for collecting all the data specified. For example, if a given Plan A is selected for the fixed
for data collection, since it may not be possible to isolate the prices for one service. It is preferable to use prices for a specific
Annex 2. ICT price data methodology 234 Annex Table 2. 1: OECD mobile-cellular low-user call distribution (2009 methodology
data usage of (a minimum of) 1 Gigabyte GB). ) For plans that limit the monthly amount
of data transferred by including data volume caps below 1 GB, the cost for the additional
bytes is added to the sub-basket. The minimum speed of a broadband connection is 256 kbit/s
Annex 2. ICT price data methodology 236 Annex Box 2. 3: Rules applied in collecting fixed-broadband Internet prices
7. The same price plan should be used for collecting all the data specified. For example, if a given Plan A is selected for the fixed
the volume of data that can be downloaded, etc 8. Prices should be collected for regular (non-promotional) plan
for price data collection, since it may not be possible to isolate the prices for one service.
GB of data volume. If providers set a limit of less than 1 GB on the amount of data that can be
transferred within a month, then the price per additional byte is added to the monthly price
so as to calculate the cost of 1 GB of data per month. Preference should be given to the most
widely used fixed (wired)- broadband technology DSL, cable, etc..The sub-basket does not include installation charges, modem prices or
-broadband price data from ITU Member States was agreed upon by the ITU Expert Group on
lessons learned from the first data collection exercise. The revised methodology was endorsed by the eleventh World Telecommunication/ICT
To capture the price of different data packages covering prepaid and postpaid services, and supported by different devices (handset and
collected for two different data thresholds, based on a set of rules (see Annex Box 2. 4
of data allowance (or validity. The customer i) continues to use the service and pays an
offers that include the minimum amount of data for each respective mobile-broadband plan The guiding idea is to base each plan on what
data allowance and validity of each respective plan Annex 2. ICT price data methodology 238
Annex Box 2. 4: Rules applied in collecting mobile-broadband prices11 1. Prices should be collected based on one of the following technologies:
plans, the plan satisfying the indicated data volume requirement should be used 8. Where operators propose different commitment periods for postpaid mobile-broadband plans,
9. Price data should be collected for the cheapest plan with a data volume allowance of a minimum of
Data volumes should refer to both upload and download data volumes. If prices are linked to †hours of use†and not to data volumes, this information should be added in a separate note
Note: ITU will most likely not be able to include these cases in a comparison
10. A validity period of 30 days should be chosen. If this is not available, 15 days should be used.
Preference should be given to packages (including a certain data volume. Pay-as-you-go offers should be used
since most often there are limits in the data volumes, either applied by throttling (limiting the speed)
of ITU supply-side indicators (i e. data collected from operators), as well as to discuss outstanding methodological issues and
EGTI is open to all ITU members and experts in the field of ICT statistics and data collection.
8 Data for fixed-telephone, mobile-cellular and fixed-broadband have been collected since 2008 through the ITU ICT Price
10 Some operators throttle speeds after the data allowance included in the base package has been reached.
Data in italics refer to ITU estimates. For further notes, see p. 248 Source: ITU World Telecommunication/ICT Indicators database
Annex 3. Statistical tables of indicators used to compute de IDI 244 Percentage of individuals using the Internet
Data in italics refer to ITU estimates. For further notes, see p. 248 Source: ITU World Telecommunication/ICT Indicators database
Annex 3. Statistical tables of indicators used to compute de IDI 246 Skills indicators Gross enrolment ratio Adult
Data in italics refer to ITU estimates Source: ITU World Telecommunication/ICT Indicators database Annex 3. Statistical tables of indicators used to compute de IDI
248 Access indicators Fixed-telephone subscriptions per 100 inhabitants, 2012 1) Incl. 524 958 WLL subscriptions. 2) Incl. payphone, excl.
change in tariff policy of the biggest WLL operator. 13) Data excluding own (NRA) consumption. 14) Excl. voice-over-IP (Voip
on 2013q3 data. 19) Refers to active Fixed Wired/Wireless lines. 20) Per June 2013.21) Operators†data. 22) Residential:
Definitive data (annual report) may change because quarterly reports use a smaller sample of operators than annual report. 24) Fixed and fixed-wireless subscriptions. 25) Excl. internal lines and WLR of
Data for the third quarter of 2013 Mobile-cellular subscriptions per 100 inhabitants, 2012 1) Numbers are down due to data cleanse. 2) ACMA Communications Report 2011-12.2) incl. payphone, excl.
Voip. 3) Active subscriptions. 4) Bhutan Telecom and Tashi Cell are the only two service providers in Bhutan. 5) Activity criteria:
voice or data communication in the last month. 6 december 2012.7) Total number of subscriptions (including non-active:
Excl. 2 720 698 prepaid cards that are used to provide Travel SIM/World Mobile service. 9) Excl. data-only SIM CARDS and M2m
Including PHS and data cards undividable. 15) Decrease was due to registration of SIMS. 16) Figure obtained from all five mobile (GSM & CDMA) operators
currently providing service in the country. 17) Incl. inactive subscriptions. 18) Preliminary. 19) Active subscriptions (87.85
Incl. inactive. 22) Incl. data-only subscriptions not possible to disaggregate the information at this point.
24) Incl. active (in the last 6 months) prepaid accounts. 25) Registered SIM CARDS (incl. inactive:
active and non-active subscriptions. 28) Incl. data dedicated subscriptions. 29) Decrease due to the closing of MTS
but excl. mobile data subscriptions (via data cards, USB modems and M2m cards). 7) Estimate
8) Incl. data-only subscriptions. 9) Data based on NRA estimates. Excl. data-only SIM CARDS and M2m cards. 10) Incl. fixed
wireless local loop (WLL) subscriptions. 11) Incl. PHS and data cards, undividable. 12) Active subscriptions (after clearing
simcards that became inactive. 13) Figure obtained from all four mobile (GSM & CDMA) operators currently providing
16 february 2013 NTA MIS. 17) Based on 2013q3 data. 18) Measured using subscriptions active in the last 90 days. 19) Per
June 2013.20) Excl. 463 646 M2m subscriptions. 21) Incl. active (in the last 6 months) prepaid accounts. 22) Registered SIM
Definitive data (annual report) may change because quarterly reports use a smaller sample of operators than annual report. 26) Data for the fourth quarter of 2013.27) Incl. data dedicated subscriptions
28) Reduction is due to implementation of sim cards registration International Internet bandwidth Bit/s per Internet user, 2012
1) Refers to a survey conducted with the following companies: Global crossing, TIWS, Embratel e Globenet. 2) This is from
MOT. 11) Data obtained from nine service operators. 12 may 2012 purchased capacity. Lit capacity: 43 096 471 Mbit/s. 13) Incoming capacity;
weekly incoming capacity, averaged over 4 weeks in December 14) Preliminary. 15) SLT Data. 16) Refers to the total capacity
4 may be revised with comprehensive data from mobile broadbnad providers. 5) Estimate. 6 june. 7) Sum of incoming capacity of all ISPS in the country. 8) Activated external
capacity. 9) By September 2013.10) Data obtained from eight service operators. 11) 1st april 2013 purchased capacity.
or laptop. 4) Data correspond to dwellings (not households). 5) Ghana Living Standards Survey 2012/2013
-Economic survey-2012.9) Census data. 10) Computer includes the number of personal computer, Notebook, and PDA
1) Labour force Survey 2013.2) Cambodia Inter-censal Population Survey. 3) Refers to PC, laptop or a tablet. 4) Data
%2) Preliminary. 3) Data correspond to dwellings (not households. 4) Ghana Living Standards Survey 2012/2013.
via mobile modem. 10) Census data. 11) Excl. households which didn†t know type of internet access 172 346 households
1) Labour force Survey 2013.2) Corresponds to all type of internet connections. 3) Data correspond to dwellings (not
instead of data survey. 13) Ghana Living Standards Survey 2012/2013. The estimate is based on weighting households who use
based on the data provided by 89.1%of operators 6) The figure is corrected. The previous figure was 1†636†700.7) Only ADSL, excl. cable modem. 8) Speeds greater than, or
than 144 kbit/s. 14) Operators data/ictqatar estimate. 15) Incl. subscriptions at downstream speeds equal to, or greater than
subscriptions. 18) Excl. corporate connections. 19) Data reflect subscriptions with associated transfer rates exceeding 200
and Fiber links. 5) Estimate, no specific data collected for â ¥ 256 kbit/s. 6) CRC estimation as of 31.12.2013.7) Estimate. 8) Data
Broadband in India with effect from 18th july 2013.12) December 13) ADSL and Leased lines. 14) Based on 2013q3 data
15) Per June 2013.16) Operators†data. 17) Incl. subscriptions at downstream speeds equal to, or greater than, 144 kbit/s (the
Definitive data (annual report may change because quarterly reports use a smaller sample of operators than annual report. 19) Estimate.
2013.20) Excl. 3175 Wimax subscriptions. 21) Excl. corporate connections. 22) 2013 data is an estimate as of June 30, 2013
Data reflect subscriptions with associated transfer rates exceeding 200 kbps in at least one direction, consistent with the
12) Satellite, BWA and active mobile subscriptions. 13) Estimate based on partial SIT data and ITU estimates. 14) Speeds
/Data refer to the sum of fixed wireless broadband and active mobile -broadband subscriptions. 16) Incl. mobile broadband and Wimax. 17) Estimate. 18) ETL and LTC. 19) Incl. narrowband
but data are not available. 23) Operators data/ictqatar estimate. 24) Refers to active mobile-broadband
effect from 18th july 2013.16) Data refer to the sum of fixed wireless broadband and active mobile-broadband subscriptions
17) 2013 data is an estimate as of June 30, 2013.18) Incl. mobile broadband and Wimax. 19) Estimate based on 1. Standard
mobile subscriptions using data services 2. Dedicated data subscriptions 3. Add on data packages. 20) Based on 2013q3
data. 21) Per June 2013.22) Mobile broadband only. Fixed wireless and satellite exist but data are not available. 23) Operatorsâ€
data. 24) As at Dec 2013.25) Q4 report. Definitive data (annual report) may change because quarterly reports use a smaller
sample of operators than annual report. 26) Wireless Broadband services are not being offered in St vincent as yet.
We anticipate that Mobile broadband and terrestrial fixed broadband services would be in place by the end of 2014.27) OFCOM
Chapter 5. The role of big data for ICT monitoring and for development 5. 1 Introduction
5. 2 Big data sources, trends and analytics 5. 3 Telecommunication data and their potential for big data analytics
5. 4 Big data from mobile telecommunications for development and for better monitoring 5. 5 Challenges and the way forward
List of references Annex 1. ICT Development Index (IDI) methodology Annex 2. ICT price data methodology
Annex 3. Statistical tables of indicators used to compute the IDI
Overtext Web Module V3.0 Alpha
Copyright Semantic-Knowledge, 1994-2011