Big data and analytics present myriad possibilities for emergency management specialists and first responders. Some of these benefits include understanding site-level impacts as they develop in real-time, adding value to programs such as Business Continuity, and creating intelligence to be used in the Planning Section of an Emergency Operations Centre.
But what exactly is Big Data? Many people think immediately of social media, and the thousands of tweets, messages, forums and other interactions that occur during emergency situations. In reality, this is one piece of a very large puzzle. Essentially, Big Data is just a fancy title for any extremely large data set that may be analyzed computationally to reveal patterns. Even more simply; spreadsheets, logs, and/or tables of data that can be analysed to produce useful information.
So based on this definition, you likely use ‘big data’ in your existing operations. If you’ve got a stack of spreadsheets hosting your Critical Infrastructure inventory, Business Continuity Plans, a long list of emergency notification contacts, or others, you’re already in the big data space.
However, when Emergency Management professionals talk about Big Data, the conversations usually focus on the many impressive case studies of its successful use in emergency situations, including applications of crowd-sourced data. One of the more commonly cited examples is the use of Big Data Tech during Super Storm Sandy, where information was used to predict the areas where assistance was most needed, and how to react to the disaster in the most efficient manner. Other examples include crisis mapping, event simulations based on real-world data, and family reunification using social media platforms. The most recent example is the presence of Facebook Safety check in the aftermath of the Las Vegas Shootings on October 2nd, 2017.
Having seen the awesome power of social media and other forms of data used for incredible good in emergencies, it is becoming more and more apparent that if resources allow for investment in this area, high-volume data can be an incredible asset. But how do we get there?
Most articles I read on this topic address questions such as:
- How can we use larger volumes of data in our programs?
- Are there tools out theme that can help us find this information?
Such queries are exciting, and I believe vital to ensure good use of data becomes the ‘new normal’ of our profession. However, these questions fail to acknowledge a core requirement for more data-savvy programs. In fact, I’d argue that this common approach represents a hidden hazard in our ability to incorporate better data management practices into Emergency Management and Business Continuity programs, and the industry as a whole.
The fundamental issue with the line of questioning above is that it assumes the existence of capacity, expertise, and a policy structure to manage the data. Beginning this conversation with how to use social media, or open data (or other common data initiatives), ignores many initial planning steps. It’s like building a house without a foundation or building codes. In essence, such projects could end up without quality standards, a common operating picture, nor any assurance of stability or sustainability. Not to mention that it overlooks the potentially huge additional workload associated with management and integration of this sort of data. Such issues are problematic even when managing small volumes of data, and do not improve with larger or more complex data sets.
Below are some common pieces of the puzzle that are overlooked when the conversation jumps directly to the use of big data:
- Spreadsheets are challenging…! Integration across multiple spreadsheets is mind-bogglingly inefficient, and managing/updating this static, isolated information is a pain.
- Corporate data governance structures are not common, and are usually contained to more data-centric business lines. Without these, no framework or guideline exists for data management.
- Neither Emergency Managers nor their colleagues in IT, communications, or other related divisions are experts in data management or analytics. This makes deeper analysis near impossible with current staffing or resourcing levels.
- Organizations with GIS or other data analysis capabilities have distinct and limited mandates that do not include assisting EM personnel with data management or assessment.
Take a typical case of Social Media use as an example. Anyone that has worked in an EOC or in an Emergency Management Program likely knows that it can be incredibly time-consuming and resource-intensive process to issue media releases and process information coming into the planning section, even if a Standard Operating Procedure, prepared statements, and templates have been created beforehand. Now add to this the vast array of potential new data sources; satellite imagery, Mapping resources, crowdsourced reports, Business Continuity Management data, social media interactions. The situation becomes immediately debilitating due to resource and knowledge limitations, before we can even think about Big Data.
Projects like Open Data are now prevalent among many public institutions, and reflect a more transparent approach to governance. Such dialogue, while clearly needed and healthy, often promises large pay-offs for the public with the “capability to distribute large amounts of data and information through many different platforms on a vast array of subjects”. However, such dialogue glosses over considerations for clear internal guidance on data collection, governance, policy or management across the organization, leaving these conversations behind closed doors and often beyond the reach of departments that have not historically produced or managed such data – like Emergency Management.
Additionally, just because the data is ‘out there’, doesn’t mean it fits the purpose, capacity, or context of the organization in question. Even if program managers are confident that data fits these criteria, keeping the information current, and ensuring its integration with current operations consists of much more than saving an excel table to a network drive, or even linking data from a live source. How do we ensure a consistent, effective and coordinated approach across sources, and how do we turn these data points into intelligence we can use?
I would be remiss if I didn’t provide some sort of mitigation for this hidden hazard. However, I should say here that I am not a data management expert; as a millennial, part-time data nerd, and Emergency Management professional I am often the ‘go-to’ for technology questions in Emergency Management settings. On that note, the first thing I’d recommend is consulting with a data Management expert! DAMA international, ISACA and a great many more organizations specialize in this area, and certify professionals for this set of expertise. Big Data is an emerging and entirely distinct professional field, and I encourage everyone to connect with these fine individuals as much as possible throughout your efforts to pursue Big Data.
That said, I will share my approach to this problem, which I implemented in order to deal with the increasing desire for data. In essence it consists of the following two core themes:
- a) A clear objective and purpose for use of the data: In essence, avoid the urge to collect all the data – be strategic and specific about the data that is needed. Rather than lofty goals like ‘predict the areas where assistance was most needed in an emergency’ – programs should begin with more attainable objectives, and ideally examine data they already hold. More on this in a bit.
- b) A governance Structure: This includes a framework for collecting, managing and using data. It should include considerations for policy, standards, data quality metrics, quality reporting, issues management, master data management, definitions, and roles and responsibilities for management of data (owners, stewards etc).
Put another way, I believe the conversation should begin with a different set of questions. Ideally, with a long, hard look at the data currently held or utilized by the Emergency Management organization.
I often use the Smart Objectives template address this. This type of questioning is of course much less exciting than thinking about how social media can change the way Emergency Management engages in response activities. However, it represents a vital and more manageable first step to approaching the use of data; building the foundation for continued program development. This framework should lead project leaders to more focused and specific questions, and a more productive way to begin the conversation. Hopefully, this can then evolve into a healthier and well-structured discussion about Big Data as your institutional understanding and capabilities mature.
Consider the following as a helpful starting point, instead of the questions highlighted before:
- What specific outcome do we wish to inform with our data?
- What attributes would be most critical to inform that outcome? (Consider running scenarios or table-top exercises to create a clear picture of potential needs).
- What data do we currently hold, or have access to within our organization?
- What is our data management structure? Are there policies, procedures or protocols for collection, use, access to or maintenance of the data? Are they applicable to our context, or do we need to create some?
- In what format is the data stored? Is this consistent across all stakeholders?
- What are the roles and responsibilities in relation to the data? Who should own, steward, control or manage the data?
- Is it currently compatible with the way we operate in an EOC/day-to-day?
- Do we have the expertise and resources use the data in an emergency situation? Day-to-day?
- Are we collecting data that is not currently used – what was the original intention of these fields? What would we need to do, to achieve this secondary objective?
- How can we improve its usability and value of existing data?
- How is it collected and maintained?
With answers and strategies in place to answer these questions, starting with only a small amount of data and clearly-defined objective, the conversation can move to capacity, resource allocation, expertise and the many other challenges. In fact, the clearer the governance structure and objectives become, the smaller these challenges seem to be.
In addition to this basic advice, below I have included some links to information I have found useful in developing my understanding of data governance, setting clear objectives for data collection, and additionally beginning to define key ‘attributes’ – the data characteristics required to achieve the specific objective.
Big data and analytics: The new paradigm in emergency management. IBM Analytics: https://www.youtube.com/watch?v=CVRmV1NYavs
Basic Concepts of Data Governance. Nicola Askham. Jul 13 2016 | 42 mins. https://www.brighttalk.com/webcast/12405/212689
Governing Big Data Panel Discussion (from Datamation.com): https://youtu.be/8haoYZn-hEI?list=LLkY2nxnWCbX4GnWuGw2yfMw
Data Management Association (DAMA) International: www.dama.org/
Data Management Association (DAMA) UK Webinars, Blogs & Surveys: http://damauk.org/webinars.php?&dx=1&ob=3&rpn=index&sid=8b5c5d393732ba17f3cd976fce454753
ISACA knowledge and insights: http://www.isaca.org/Knowledge-Center/Pages/default.aspx
Data Governance: Priority in a Data-Driven World: Creating a roadmap to protect and manage vital data assets: http://strategyonline.ca/2017/07/18/data-governance-priority-in-a-data-driven-world/