Transcript: Government of Canada Data Conference 2023: Agriculture and Agri-Food Canada's COVID-19 Dashboard Project
[Elise Legendre and Warren Goodlet sitting in two chairs turned 45 degrees to the camera]
Hi, I'm Elise Legendre. I am the Chief Data Officer for Agriculture and Agri-Food Canada.
And I'm Warren Goodlet, Director General of the Research and Analysis Directorate at Agriculture and Agri-Food Canada.
We're here today to talk about a joint data effort between our two teams.
At the onset of the Covid-19 pandemic, senior management at AAFC wanted to have the most current data available to be able to understand how the pandemic was impacting the sector.
Given the importance of agriculture and agri-food to Canadians, and the numerous supply chains involved, a weekly PowerPoint dashboard was developed to brief the senior management team on key issues and bottlenecks evolving in the sector using a broad set of information and agricultural indicators. This dashboard built in PowerPoint met our needs and provided senior management with key information.
Given the wildly dynamic situation early in the pandemic, the list of indicators we used and how they were presented changed frequently.
But at two years into the pandemic (last March), the process had evolved and a number of the inputs to the dashboard became more regular data indicators. For example, thing like food retail sales, inflation, US indicators, Canadian price info, and a number of other metrics were included in every edition of the presentation. So we started to ask if there was a way to reduce the labour required to produce these products every couple of weeks.
This sparked us to investigate the potential for automation.
That's when my team approached Elise's team to see if there was another way to go about this.
My team saw an opportunity to automate many of the tasks that analysts needed to perform to update the dashboard.
We saw it as a great test case for harnessing data across and outside AAFC.
In this particular case, analysts were manually downloading data from close to 40 different data sets, including Statistics Canada, the Bank of Canada, and other government departments.
They then used that data to create static graphs and charts, which they would insert into the PowerPoint document.
So the first thing we looked at was moving the dashboard into Power BI so that we could take advantage of its interactive features.
We wanted the Power BI dashboard to connect directly to the data sets so that it could update the graphs and charts automatically.
This would give the analysts space to focus on higher-value tasks, like providing their expert analysis and insights.
We collected data from open source, like Statistics Canada, the Bank of Canada and a number of U.S. government agencies.
When the data was in a format that wasn't east to automate, we scraped from reports from federal departments and industry partners.
We also extracted data stored in various internal databases at AAFC.
Our teams worked closely together for more than six months. Elise's team needed to understand how we got our data and performed our calculations to create our dashboard. And my team got exposed to new tools and ways to work with data.
It was a good learning experience – I think our teams learned a lot from each other about what goes into creating a product like this.
I think this experience also showed that data and automation is as much about technology as it is about people and how we work together.
We needed to shift how we collaborated and thought about data to be more inclusive and efficient.
It has also put a very tangible spotlight on the importance of data management and data governance and how, if it is done right, we can be more efficient. Analysts that were involved in the project have reported that it now takes them 70% less time to produce their input for the dashboard – that's is significant!
The end result will be a highly interactive, fully automated dashboard. We have a few things to work out but we are almost there.
Given this dashboarding project has integrated many experts and resources from across the department, it has been a unique and evolving opportunity to determine how to balance the benefits of having experts brief on key evolving situations in the sector while finding ways to reduce the time needed to keep data and visuals up to date in a dashboard.
It's also made us think about things like information management – how do we get approval to move up a briefing if the information in it keeps being update? What are the mechanisms we need to put in place to control versions and limit confusion? It has made us reflect on the way we work and has the power to really transform how we share information.
It is early days, but we are hoping the synergies will be significant and that the lessons learned will be broadly spread given the broad participation in the project, and that analysts will consider automation in other tasks they routinely perform.
The virtual environment has allowed for a broad group of experts to run through particular elements of the dashboard and delve deep into their areas of expertise in a relatively short time. Approaches that can re-balance the efforts on insights versus the wrangling of data are helpful.
It was also a great test case for future efforts.
One of the most important things we took away from this work was just how important it is to have buy-in at all levels of the organization for a project of this scope to succeed.
It also showed us that we still have work to do to raise our level of data awareness so that people know what data is available and how it can be accessed.
Overall, we're proud of this work and I think it demonstrated how much we can accomplish when we work together.
And we're looking forward to working together on many more projects.