Home Blogs Cracking the data before the crisis

Cracking the data before the crisis

4 Minutes Read

The humanitarian sector is shifting away from a historical emphasis on operations and logistics towards use of data science and predictive analytics to forecast, prepare for and prevent emergencies. In 2019, Mark Lowcock, the Head of the UN’s Office for the Coordination of Humanitarian Affairs (OCHA), said “Most of the time when humanitarian agencies are responding to crisis and disaster, they are reacting to things that have happened…The challenge is to look ahead and to enable us to predict and anticipate, so that we get a response which is faster and cheaper and reduces suffering.”

Humanitarians are a little behind the curve compared to other industries such as finance and insurance. This is perhaps unsurprising given they often don’t have access to the technical development resources available to those early adopters. The way the sector has historically been funded, too, can inhibit innovation. With most grants based around specific programmes designed to deliver concrete humanitarian outcomes, donor organisations have understandably tended to favour tried-and-tested methods over experimental schemes.

However, there has been a growing impetus in recent years – backed by significant grant funding – to innovate and apply data science to understand and tackle huge and growing crises such as the impacts of climate change. This coincides with an increase in the availability of data, rapid advances in technology and a widening base of people in the sector with the skills to apply it.

Insight from predictive analytics is now being used to improve humanitarian supply chains and reduce costs as well as to forecast things like food shortages and movements of displaced people. UNHCR, the World Bank, the Red Cross and Save the Children are among the organisations actively applying predictive analytics to solve these types of problems.

According to a recent Institute of Development Studies report documenting 49 humanitarian predictive analytics projects, “Historic data of previous humanitarian events plus mobile phone records and social media posts can provide the high volumes of data needed to analyse food security, predict malnutrition and inform aid deployment. Satellite images, meteorological data and financial transactions can be used to track and predict the escalation and trajectory of refugee movements.”

The COVID effect

Humanitarian data science predates COVID-19, but has been precipitated by it. The pandemic has significantly increased humanitarian need globally in a short timeframe. At the same time, funding is predicted to contract as economic impacts deepen. Meanwhile, reaching affected communities is harder due to infection risk, social distancing and travel bans. What’s more, humanitarians themselves are directly affected, as well as the communities they serve, which has never happened before. In short, COVID has made humanitarian work even more difficult and complex, necessitating smarter, more agile approaches.

According to Sarah Telford, Centre Lead for the UN OCHA-run Centre for Humanitarian Data in The Hague, “The COVID-19 pandemic has brought into stark focus the need for data and the value of models to inform response strategies. Anticipatory action is no longer an abstract idea but something we are all doing by staying home and increasing hospital beds.”

Since late 2019, the Centre has been actively working on predictive analytics in order to enable humanitarians to anticipate emergencies, forecast their effects and trigger responses earlier.

One of the projects the Centre’s predictive analytics team is working on, in partnership with the John Hopkins University Applied Physics Laboratory, humanitarian geospatial agency MapAction and individual OCHA country offices, is the development of COVID-19 modelling tailored for country-specific contexts. Using data from aid agencies, the Humanitarian Data Exchange and WorldPop, this seeks to predict the scale, severity and duration of the outbreak within each country, including its likely effects on particularly vulnerable groups, such as people at risk of hunger or using solid fuel indoors for cooking, and the effects of non-pharmaceutical interventions like travel bans and face masks. Because it looks at projections for specific vulnerable groups, as well as the general population, at a sub national level, this should be particularly helpful for COVID planning.

Also Read: Tracking and understanding the impact of humans on environment

Automation

Usually when crises happen, a ‘data scramble’ occurs to rapidly collate the best-available data to generate reference maps that humanitarian teams need straight away as the foundation blocks upon which to design their response. As the situation evolves, layers of situational information are added to these baseline maps to communicate aspects like the types, scale and severity of humanitarian needs in different areas, who is doing what to help and where.

MapAction is now also automating the production of core maps and geospatial products known to be needed from the first stages onwards in different types of emergency. Automating the production of these core maps will therefore enable the faster creation of a range of further maps showing what is happening on the ground. They will initially be available for twenty countries identified as being highly vulnerable to humanitarian disasters, in a project funded by the German Federal Foreign Office.

Doing this in advance of a crisis has great advantages, as well as solving some fundamental data issues affecting the whole humanitarian sector.

“In each case, we’re either looking to answer a specific question that humanitarians need to answer, or look at the generic elements that are fulfilled by data, and which could therefore be automated,” explained Juan Duarte, MapAction’s Technical Director. “We want to shift away from spending time on repetitive tasks. This will free up humanitarian GIS specialists to focus on the unique aspects of the situation they’re dealing with.”

Sourcing good data and identifying gaps or anomalies is the biggest – and arguably most valuable – part of this 18-month project. “Finding data gaps is like looking for a missing piece of a jigsaw puzzle,” Duarte explained. “When a map doesn’t meet our standards, it’s because a puzzle piece is missing. Since we know the puzzle well, having solved it many times, we can identify the right piece, or the shape of the piece when it’s missing and needs creating.”

MapAction is working with humanitarian and disaster management agencies around the world to rectify identified data gaps. The aim is to develop an approach that will be applicable for many years to come, with consistent, documented processes.

Collaboration is key to maximising the potential benefits of humanitarian data science. It’s important that those leading the field stay in listening and learning mode and ensure that smaller, local actors, which may themselves lack the necessary skills and resources, are not excluded from participating. Similarly, we must stay cognisant of the errors and prejudices that may be inherent in historical data as well as the safeguarding issues associated with capturing and processing data.

Otherwise, there is a danger that new practices may simply replicate and even amplify existing dependencies and inequalities or cause active harm. Playing catch-up to the private sector may not be a bad thing, as it means humanitarians can avoid some of the pitfalls which could have negative outcomes for people caught up in crises, and it may increase the speed of travel, so more lives can be protected, sooner.

Also Read: Location-aware leadership! Does it really matter?

Previous articleIndian Government rolls out GIS-based land bank platform 
Next articleDeadline extended for MyGalileoDrone competition
Avatar
Monica Turner is a humanitarian data scientist at MapAction, based at the UN's Centre for Humanitarian Data in the Hague. In 2015 she completed her PhD in astrophysics, where she focused on the statistical analysis of distant, bright galaxy spectra to determine the composition of gas in the Universe. She then worked at Las Cumbres Observatory where she built pipelines for processing astronomical data. Before starting at MapAction, she spent 6 months volunteering as a data scientist at 510 global (part of the Netherlands Red Cross), where she worked on automating the assessment of building damage due to natural disasters.