C O V I D 19

Data Story

COVID-19 Observatory - Opportunities and Challenges

The objective of the COVID-19 Big Data Observatory is to raise the awareness, access, and ability of local policy makers to effectively leverage emerging technologies and information sources to support better public action design and implementation. We group key prospective insights and techniques into five major categories, guided by the nature of the data and the types of approaches needed to distill actionable insights. Depending on the disclosure and privacy policies associated with each data resource, what can be done differ. In terms of applied insights, however, the most critical factor will be whether the respective resources can actually be accessed in practice in a timely manner through functional “data pipelines” – especially through Application Programming Interfaces (APIs) - to generate value.

  • Human Mobility
    Data
  • Satellite
    Data
  • Crowd-sourcing
  • Transactions
    Systems
  • Text Mining


Human Mobility Data


Whether by way of basic phones or smartphones, the movement of people is now captured by billions of electronic devices. Additionally, many vehicles are now equipped with location trackers. Various fixed-point sensors, for example CCTV cameras, can also capture the movement of vehicles and people. Mobility data is incredibly valuable for understanding how effective different policies to contain the spread of the virus are by specific locations (e.g., social distancing, lock-downs, etc.). Data is derived from mobile phone Call Data Records (CDR) or smartphone applications, as well as mobile or fixed sensors. During the recovery phases, further tracking of these data can also understand the degree to which mobility is returning to “new normals.”


Satellite Data


Earth imagery can complement mobility data by similarly mapping ground-based activity at high levels of spatial and timely granularity. Basic visual narratives provide a compelling insight into visible changes to even changes in traffic on roads. The use of satellite nightlights data has found some general use for now-casting of sub-national GDP proxy. Satellite data can also be used to monitor environmental emissions, which in many locations have declined with COVID-19 measures. But better nightlights data is now also offering opportunities to analyze change at a neighborhood or road level, for example in Wuhan, China after the initial lock-down in January 2020. Satellites can also help in the remote monitoring and supervision of COVID-19 public infrastructure developments, for example as illustrated by the building of a temporary emergency hospital in Wuhan. Since informal settlements or slum areas can be especially vulnerable to COVID-19, satellite imagery can help quick map and validate these likely hot or “stress” spot areas.


Crowd-sourcing


Smartphones or tablets now represent an incredibly powerful way of rapidly collecting or aggregating frontline data. This can be done on the basis of directed surveys, either by government authorities or global platforms. COVID-19 awareness and impact tracking by PREMISE relies on hundreds of thousands of paid “micro-task” contributors across the world. Platforms such as Open Street Maps can help rapidly map the infrastructure available to vulnerable groups. A variety of crowd-sourcing mechanisms are being used to both compile lists of COVID-19 related developments and actions across the globe, but also to suggest innovative solutions to address the crisis.


Transaction Systems


The digital transformation of economies and governments across the world means that a wealth of highly granular data now exists to monitor the impact of COVID-19. Analysis of on-line shopping patterns, use of digital on-line public services, digital payments logs, government payments and procurements, or airlines bookings and flights can all provide near real time insights into the disruptions and prospective rebounds in the current crisis.


Text Mining


The COVID-19 crisis has seen a massive continuous increase in text materials. These range for newspaper articles, academic publications, official websites of national governments and international organizations, as well social media posting. The challenge for applied text mining is to deliver results that can be readily used to meaningfully inform different stakeholders potentially overwhelmed by text inputs. Part efforts to address a possible infordemic - too much data of sometimes questionable quality - is to monitor for such concerns as fake news. The power of topically and spatially granular text mining comes from being able to rapidly overlay this with data from other sources. For example, if the mobility data shows that lock-down is working effectively, but sentiment is rapidly turning, this might need to be an early warning indicator for governments. Or text mining filters can be used go hone in on particular topics (e.g., hotspots in terms of particular areas, rule breaking, or issues not yet reported in mainstream sources).

Feedback

Know of other specific examples of where big data is helping inform COVID-19 public action in a clear, accessible, but above all tangible manner, let us know.