Leveraging Mobile Network Big Data for Developmental Policy: Research papers from the project

"Many developing countries lack the capacity and resources to collect and analyze data for evidence-based policy-making. Is big data, which involves large and complex data sets, an opportunity to meet this challenge? Or will it become difficult for developing countries to adopt for solving problems?" - IDRC

The Sri Lankan think tank Learning Initiatives on Reforms for Network Economies Asia (LIRNEasia) has been exploring the possibility of using big data to inform public policy since 2012. Supported by the International Development Research Centre (IDRC), this research focused on transportation planning in urban centres in Colombo to better integrate different parts of cities and suburbs. LIRNEasia is using mobile network big data (MNBD) to: improve transportation planning in urban centres in Colombo, model the spread of infectious diseases, and uncover fine-grained indicators of economic activity.

Some argue that anonymised big data can offer important advantages in generating evidence and offer new insights into solving social and economic problems. Mobile phone data (one form of big data) in particular has characteristics that make it valuable for informing public policy in developing countries. First, as many, if not most, citizens now carry a mobile phone, the phones themselves act like sensors that yield data as a by-product. As such, mobile phone data is the only kind of data that comes close to comprehensive coverage of populations. Second, since people carry their mobiles as they move around the city during the work day, this produces important data on movement patterns which can help with better urban planning.

As one of the few think tanks in the global South conducting big data research, LIRNEasia has undertaken work to frame the debates emerging around the viability and use of large data sets. In its research to date, LIRNEasia has reached some general conclusions that can inform the next generation of big data research:

• In this very nascent field, experts are scarce, particularly in the global South, and more intermediaries like LIRNEasia are necessary;
• There is a large gap between these huge data sets and their accessibility, usability, and usefulness;
• Trust is the foundation for the cooperation needed to free private sector data for the public interest;
• There is no standard guideline or framework for ethical research in big data for development; and
• Replicating research is a learning process - among a wide variety of information ecosystems in diverse national contexts, there are too many factors at play to assume there is sufficient knowledge to replicate "success".

The central problem addressed in this research project is how to realise the potential of big data to generate evidence to better inform public policy. But it also includes developing safeguards that will ensure the responsible use of data for the public interest and understanding what is involved in developing capacity among researchers, technical staff in national statistical offices, those affiliated with local universities, and among the mobile operators themselves.

As the research is now currently extending to other countries in the region, including Bangladesh, project organisers assert that it stands to have a significant impact on transportation systems in major urban areas. It could, for instance, allow Colombo and Dhaka (Bangladesh) to become more efficient in dealing with traffic congestion, pollution, and waste management.

Many research papers have emerged from this project thus far, several of which can be found below. These were presented at the Communication Policy Research South (CPRsouth) conference in Taipei (Taiwan) in August 2015. This annual conference aims to build policy research capacity in the South by reinforcing the knowledge, research, and presentation skills of scholars from around the global South working in information and communication technology (ICT)-related policy.

1. Where did you come from? Where did you go? Robust policy relevant evidence from mobile network big data by Danaja Maldeniya, Amal Kumarage, Sriganesh Lokanathan, Gabriel Kreindler, and Kaushalya Madhawa, March 2015 - Abstract: As people become increasingly mobile and the urban traffic patterns become more complex there is an emerging need for the transport planning to become a truly continuous process. Developing countries that continue to rely mainly on infrequent and expensive survey data for transportation planning and management, will struggle to cope with this changing dynamic. Mobile network big data provides a promising avenue that can potentially fill this gap without requiring significant investment in sensor networks. This paper builds on earlier work to explore in greater depth the potential of mobile network big data (MNBD) to provide high value mobility insights that can support a continuous approach to transport planning. We evaluate the levels of accuracy and detail that mobility insights based MNBD can deliver by comparing multiple recently developed approaches for estimating mobility and validating the results against the data from traditional transport forecasting methods. We discuss inherent limitations of MNBD in generating insights for transport planning and propose various methods to address these limitations in future work. The value of such work is that this extends the state of the art of using such new data sources that can transform traditional transportation planning.
2. Using mobile network big data for land use classification: CPRsouth 2015 by Kaushalya Madhawa, Sriganesh Lokanathan, Danaja Maldeniya, and Rohan Samarajiva, July 2015 - Abstract: Understanding and monitoring land use characteristics is critical for urban planning. Unfortunately the traditional way of generating insights on land use involve surveys and censuses, which are both infrequent as well as costly. In developed economies that have greater levels of datafication, as well as use of remote sensors, the dependence on traditional methods is waning. However developing economies continue to depend on these traditional methods. Mobile phone use however is nearly ubiquitous even in developing economies. This enables the population to in-effect act as sensors of human activity. This paper explores the potential of leveraging massive amounts of human mobile phone usage data (i.e. mobile phone big data), to understand the spatiotemporal activity of the masses, and by extension provide a useful proxy for activity-based classification of land use. Using unsupervised clustering techniques on the mobile phone activity signatures of the population aggregated at the base station level, the paper shows the feasibility of inferring at the very least three distinct land use characteristics: commercial/economic, residential, and mixed-use.
3. Understanding communities using mobile network big data: CPRsouth 2015, by Kaushalya Madhawa, Sriganesh Lokanathan, Rohan Samarajiva, and Danaja Maldeniya, July 2015 - Abstract: The patterns of human interactions are not random, but intertwined with many attributes of individuals such as ethnicity, economic status etc. Using anonymized call detail records obtained from a mobile operator in Sri Lanka, this paper investigates the communities formed by the communication patterns. By applying several community detection algorithms, we could identify a community structure consisting of 11 communities as the most suitable one. These communities show similarity to the nine provincial boundaries at varying degrees. But all these communities show high level of spatial coherence. Additionally we explore how these communities segment into a further level of sub-communities.

LIRNEasia's research is being disseminated through online forums, conferences, lectures, and person-to-person exchanges. LIRNEasia regularly meets with policymakers around the world to discuss their research. In Colombo, they participate in a Big Data Meet-Up associated with the project to explore big data for development in Sri Lanka. And among other speaking engagements, the Team Leader of the Big Data Research project, Sriganesh Lokanathan, gives talks around the world, including in Ottawa (Canada), Brussels (Belgium), and New Delhi (India) [as part of the IDRC distinguished lecture series there].