Tableau: DSI's not-so-secret decoder ring
IU's Decision Support Initiative (DSI) uses Tableau visualization software to help turn a bunch of silent numbers into a message that can lead to all sorts of things, from found money to more precise faculty hires. Like a secret decoder ring, Tableau can help decision makers see what all that data means, so they can act on it with confidence. The better we are at using Tableau, the clearer the message becomes. To that end, DSI held its first-ever hackathon.
An invitation-only event sponsored by Tableau, the DSI team brought together people from across campuses, colleges, and disciplines to work with a dataset and see which team came up with the best visualization. Teams were intentionally crafted from functional experts (i.e., data people) and IT professionals (i.e., Tableau gurus), the idea being that, as in real-life situations, these two types of people need to work closely to arrive at the best solution for visualizing and easily interpreting data. They were then given a dataset and a user story, which outlined the scenario, including who would be using the data, how the data would be used, and what big decisions needed to be made using the data. Then they set to work to try to make the best visualization possible in just a few hours.
"What we're looking for are ways that we can extend our capabilities," said Dennis Groth, vice provost for undergraduate education in Bloomington. "Ways that we can bring other people up to speed in the community faster. How we can learn from someone across campus that they've already solved a problem in a particular way, so we don't have to reinvent the wheel. We can actually move the institution ahead and move faster, be more nimble, develop better solutions for decision makers."
Because the hackathon was an exercise and not the real deal, complete accuracy was not required or expected in such a short time frame, but even then, it wasn't an easy task to squish into a few hours. Figuring out how to ask the right questions of the data, and how to coax that data into something university decision makers can see and understand clearly, were two of the toughest hurdles participants faced.
"The biggest challenge is the business question," said software engineer Azadeh Sanjari, whose team won first prize for best visualization at the hackathon. "Having the right question on a dataset is the most important thing. So you may be familiar with the data, you may know all the techniques that you want to work with data, but it's very important to ask the right question."
For example, a client may say, “I need to see the number of employees by department by gender over the past 10 years to better understand the gender makeup of our workforce.” While it's easy to display this in Tableau, the report will simply show gender breakdown with filters by department, requiring the user to look at each department, one by one, and interpret the data to help drive a decision. For the purposes of DSI, a better way to word this request might be, “I need to understand our workforce gender distribution, identifying where we are doing well relative to university targets and where we need to improve, so that we can focus our recruiting efforts appropriately.” This would more likely drive the visualization towards showing outlier departments with low levels of diversity so the user doesn’t have to search for them.
Tools like Tableau make sophisticated visualizations a possibility for anyone. You don't have to understand the composition of an operational data store or a data warehouse and its relationship to the production data store. You don't have to understand how to extract, transform, and load the data. Tableau does it all for you.
"Tableau really has made a stake in the industry as not a technical person's tool," said Rob Lowden, associate vice president for enterprise systems. "The tool makes it very simple to get in, take disparate sets of data, and visualize them in ways that didn't previously exist."