Companies are increasingly relying on business analytics to extract value from the large volumes of computer-readable and analyzable (or “datafied”) data in their possession. Big data for development (BD4D) seeks to apply these techniques to big data held by both government and private entities to answer development-related questions. Given low levels of “datafication” of transactions and records in developing countries, analysis of credit-card use or even social-media use is unlikely to yield coverage approaching n=all as in developed countries. Mobile transaction-generated data (including Call Detail Records or CDRs) are an exception. Because they can yield information on movement of people, they have great potential to inform a host of policy domains: urban and transportation planning, health policy by enabling the modeling of the spread of infectious diseases, socio-economic monitoring, etc.
Yet the public discourse on using mobile network big data for development can be characterized as a competition of imaginations between hype and pessimism. The practical challenges and opportunities of obtaining, analyzing and taking evidence from such Big Data to policy are buried in the fog of such imaginations. This session will tackle 3 issues:
• Access to such data has been the domain of relatively few researchers. Such access has been preceded by long negotiations and bound by lengthy agreements. This raises questions regarding how the transaction cost of negotiating access and extracting data can be reduced, whilst safeguarding privacy and competitively sensitive information.
• Similarly as evidenced by the recent disclosures regarding the declining veracity of Google Flu Trends, the state of the art in analyzing and determining the veracity of such big data analyses is still evolving. The statistical challenges of big data determine their utility and applicability for policy.
• Even when policy-relevant results are obtained, taking such evidence to policy makers is not easy. Knowledge of big data and its associated techniques are limited amongst policy makers and the broader symbolic environment. The results from big data analyses at times challenge the extant well-established paradigms in generating policy relevant evidence.
Joshua Blumenstock and Sriganesh Lokanathan presented findings from big data research likely to be of interest to policymakers. Professor Amal Kumarage of the University of Moratuwa (and also Chair of the National Transportation Commission of Sri Lanka) and Mr Ratnapriya Wickramasinghe, Assistant Director at the Sri Lanka Department of Census and Statistics provided the perspective of the policy makers. The session was moderated by Rohan Samarajiva. The audience was not large (<30) but fully engaged. There were multiple rounds of questions and answers.