Saturday, December 6, 2014

Big data: mobile data for development


When I said in my last post that 'One that will be interesting to the general public, administrators, or researchers, and "must read" for MPT, Ooredoo, and Telenor our mobile communication providers if they haven't done so is the Mobile Data for Development Primer by Global Pulse ...', for the must read part I had in mind the young local staff and not those of managerial or professional levels, or the expatriates. I suppose ordinary folks could never be in their league, at least for the time being.

However, when I looked up just now about Ooredoo and Telenor in Myanmar they seem to have been doing something more than providing telecom services. I found Ooredoo launching Myanmar Maternity application “maymay” and Telenor launching "Telenor Light Houses" (Community Information Centers).

The figure below is from 'Using mobile data for development' report available at:
It is good to see that Telenor has been active in the use of mobile data for epidemic surveillance and disease containment and information strategies.

We have seen that Mobile Phone Network Data for Development published in November 2013 by Global Pulse is a primer on how analysis of Call Detail Records (CDRs) can provide information for humanitarian and development purposes. It is essentially an advocacy document for the utilization of metadata from mobile phone calls and messaging for development. In a nutshell, "the document explains three types of indicators that can be extracted through analysis of CDRs (mobility, social interaction and economic activity), includes a synthesis of several research examples and a summary of privacy protection considerations".

The 'Using mobile data for development' report, produced by the Bill & Melinda Gates Foundation in conjunction with strategy consulting firm Cartesian and published in May 2014, covered the same themes as the Global Pulse Primer, but in considerably more detail. Additionally the first two chapters give (i) the situation of the adoption and usage of mobile phone in the developing world, and (ii) what data are captured by the mobile data systems and how they could be interpreted and used. These would be most useful to analysts and researchers.

The table of contents of this report gives a good idea of what to expect:

1.  Executive summary
2.  Adoption and Usage of Mobile Phones in Developing Countries
         Mobile Adoption Rates Are High in Developing Countries 
         Rates of Mobile Ownership and Usage Are Equally High among the Poor 
         Usage Patterns in Developing Countries Differ from Those in the Developed World

The report says that people in developing countries use more voice and SMS text as compared to data. However in Myanmar because of lack of alternatives most of us have to use mobile phone for internet connection and so our data usage could be markedly higher than others. I guess most of us who own a desktop or a laptop would be using smart phones as wifi hotspots to connect. Before mobiles and cheap SIM cards became available, I used the slower than slow prepaid dial-up connection at home which was useless for purposes other than reading email. If I need to visit the Internet cafes, their speed was not too much faster. Currently the Wimax and other options are too expensive. While ADSL by MPT is quite good it is not widely available. I lived in downtown Yangon and yet when I applied for ADSL, I was told that it is not available for the whole of my township because some kind of equipment or facility is not there.

The fact is that if we can't have some reasonably priced broadband option like cable networks becoming available, we would still be with a virtual connectivity drought. Meanwhile, MPT, Ooredoo and Telenor should make the data rates cheaper too and what is better than letting the market take care of it, for example, by promoting cable companies and startups.

For a people denied of the advantage of good and cheap mobile communication until recently, we would like to see how we compared with others. Unfortunately we aren't there.

I don't know if there is good data on that for Myanmar. Anyway, from the wiki page 'Demographics of Burma' we take the percent of population 15 years and over of 72.5% as the proportion of adult population (consistent with figure-1) and using a growth rate of 1.07% and the total population as of 2014 census at 51,419,420 we calculated the 2012 population. With the estimate of number of mobile phones as 5,400,000 for 2012 (wiki page, Telecommunications in Burma; which is quite rough) we get an estimate of ownership among adults of about 15%.

3.  What Is Captured in Mobile Data Systems? 
         How a Mobile Network Functions 
         Mobile Data Captures a Wide Range of Customer Behaviors 

Under this topic the three most important type of information captured are:
(i) "Location and mobility: Location is tracked passively when users’ phones interact with towers in each cell cite they visit, and actively, each time a user initiates a voice call, SMS, or other transaction."
(ii) "The social network: Calling and SMS patterns create a lens into a person’s social network, including who they communicate with, how long, and how often. Further, in many emerging markets the originator of a phone call pays for the minutes of the call; even understanding who someone calls vs. who calls them can give a sense of social stature among a social network, important for marketers seeking to reach nodal hubs of influencers."
(iii) "Recharge and purchase history: Patterns of recharging minutes and purchases of VAS (value-added services) can give insights into an individual’s economic circumstances and the financial shocks or difficulties they face."

         How Mobile Data Is Captured From User Interactions 
         How Mobile Data Can Determine a Person’s Location 
         Gauging the Availability and Accessibility of Data 
         Understanding and Interpreting Different Kinds of Mobile Data 

There is great potential in that "signal characteristics such as attenuation and signal distortion can be measured to provide an indication of local ecology, rainfall patterns, civil construction, etc. Figure 11 explains how attenuation from radio signals was used to collect a large amount of accurate and timely rainfall data."

4.  Present and Possible Applications 
         Potential Insights from Mobile Data 


  
         Mobile Operators Are Beginning to Exploit Mobile Data 
         Current Uses of Mobile Data in Development Programs 
         Future Opportunities to Leverage Mobile Data 
         Should Mobile Data for Philanthropic Use Be Free? 

"The United Nations Global Pulse has put forward the idea of 'data philanthropy,' where operators would have a duty to share data for certain limited uses when the public good is urgent and clear. Global Pulse argues that these cases actually make business sense ... "

5.  Regulatory Landscape and Data Privacy Considerations 
         Regulatory Regimes Are Becoming Clearer and More Standardized 
6.  Considerations for Data Sharing 
         Sensitivities around Mobile Data Access 
         Commercial Sensitivities around Data Access 
         Public Opinion Plays a Role 
         Approaches to Protecting Data Privacy 
7.  Conclusion

"Mobile data has enormous potential to support development efforts and through this to improve the lives of poor people around the world. ... Mobile data offers larger and more representative samples, in near real-time, and at far lower costs than alternative means of data gathering. Indeed we believe the opportunities to leverage these data sets for development goals are only starting to be explored. ... There is an opportunity to learn from the best examples that have been demonstrated to develop a foundation for broader use of mobile data ... ".

It seems clear that mobile CDR data has enormous potential for development applications. To enrich and accelerate research efforts, third party and government data sets will have to be combined with those of the mobile CDR data. That means those areas can't be neglected and they need to develop along with big data. To be able to generate useful CDR data, mobile communication service penetration into the population and rural areas have to be sufficiently high, though urban related, or some other applications may be feasible earlier. At present, big data requires big hardware and perhaps big brains and to round them up—big money. So, it will again be the old story: when some opportunity opens up, a handful of people who could take advantage of the situation would take all. But when you are lagging behind and you want to leapfrog, you would need more than the critical mass of smart people needed just to keep the machines running.


No comments:

Post a Comment