Data Analytics
Mining Data Across the Campus
An increased focus on data at Clemson University is impacting decision-making in finance, operations, facilities and more.
At Clemson University (SC), data architects and business teams are developing a multidimensional database about campus facilities. By merging in data from financial systems, Brett Dalton, vice president for finance and operations, can see the size of rooms and their condition and can ask questions about the university's costs associated with them. "We are creating a robust picture of an asset, how it is managed and what it produces in terms of bottom-line productivity," he said. "We can be more strategic in targeting preventative maintenance dollars and renovation dollars. Some things come to life when you look at them in an integrated fashion. We might decide we need to tear a building down."
That approach is an example of Clemson's increasing focus on data-driven decision-making, which is spreading from finance and operations to other parts of the university, including facilities, research and even sports. Dalton said he is not certain the desire or need for analytics has changed, "but because of the tools and expertise and the continuously dropping cost of doing this type of work, we are able to utilize data-driven decision making more readily, more freely and across a wider spectrum. The big thing is that the tools and technology available make it so that if we fail to utilize them, we are being irresponsible."
Dalton, who has responsibility in the area of strategic planning and implementation of the school's 10-year plan, said the approach taken in finance has opened a lot of eyes across the university. "We are setting a new standard that says we are not going to let you make bad decisions or ones that haven't been thoroughly vetted and analyzed. I think we have raised the bar."
Making Data-Informed Decisions
Matt Chambers, a data architect, uses data visualization to help his institution's leadership communicate about strategic decisions. He also works designing the data platforms that facilitate data analysis. He said the effort touches a lot of different types of data: financial, human resources, space and facilities planning and student enrollment. "We are responsible for predicting the enrollment of the student body and associated revenue from that. We are building predictive models and using Tableau [business intelligence software] to monitor the performance of the models," he said. One dashboard queries the student information system every night and pulls in enrollment data and billing data. It overlays Clemson's prediction models on top of that so that analysts can see how accurate they are.
For instance, Chambers said, "When the semester begins, we might show more revenue than a couple of months in, once everyone has dropped classes. We can see that revenue fluctuation day by day, and we can adjust our model's performance to compensate for which students are dropping classes — and we can make sure our model is accounting for all that."
Chambers is leading a grassroots initiative to get more departments using Tableau as a data visualization tool. "We have brought other divisions onto our Tableau server," he said. "We have human resources and the comptroller's office, and we are working with areas such as athletics and the research division. It hasn't become an enterprise-wide initiative yet, so it is still kind of in our shop right now. When it becomes an enterprise-level thing, I think IT will manage some of the operations. That would free up more of my time to do analytics and visualization work."
Improved Campus Planning and Budgeting
Besides long-term strategic planning on campus, another effort involves shorter-term lecture hall utilization to help campus-planning executives see buildings and classrooms that might be underutilized. "We have dashboards where you can see the entire campus laid out and you can see the buildings colored by their utilizations," Chambers explained. "When you click on a building, you get a tree-map view of all the rooms sized by square footage and colored by utilization." Click on a room and you can see its utilization throughout the day based on time blocks. Users can filter by different types of rooms: They can look at auditoriums, just small classrooms or rooms that have better technology. They also can see how policy changes and scheduling can improve utilization from semester to semester.
Chambers said Clemson is working toward a comprehensive suite of campus planning applications. "We have to be sure the space we have on campus can support whatever growth initiatives enrollment management has in mind," he said. "We are not there yet, but we are trying to get to where we have a predictive model for space utilization as well. Based on the inputs that go into the enrollment strategy and enrollment predictions that come out of that, we can push that into a space utilization model that would show us if we have a specific recruiting target in mind, are we going to be over-allocated on space utilization in certain buildings, or do we need to optimize that to take advantage of space we have available, or do we need to justify building a new facility."
Dalton noted that another recent change involves how his division gets data to the board of trustees, which meets quarterly. It used to be that board members would receive printed quarterly budget reports. "We don't even bother sending them those anymore," Dalton said. As soon as a fiscal month closes, there is an e-mail reminder that goes out to the board with a link. "Instead of providing them static information four times a year, they are able to see it in a more timely manner," he said. "We can push more content to them and it is more interactive. If they have a question, they can dig more deeply into a certain trend. It frees us up to take them more deeply into the operations."
The uptake of Tableau as a data visualization tool in other departments is happening gradually. "We have a higher percentage [of users] now who are just consuming content vs. being content creators, but part of what I have been doing is the evangelism about what they can do with the tools and how easy it is," Chambers said. "If you can use a pivot table in Excel, you can use pretty much any Tableau dashboard. That is an easy frame of reference for a lot of people. But we are still in our infancy as far as growing the user base of people interacting with it."
Open to Misinterpretation
As data-driven decision-making spreads on campus, there is a potential downside, Dalton said. "Just because you have access to a tool doesn't mean you are capable of utilizing it," he said. People can run into problems if the questions are not properly formulated or if the data are not carefully managed and interpreted.
"Tableau is a phenomenal tool, but it is primarily a visualization tool," he noted. "Just because it looks like there is a relationship between variables doesn't mean there is one. Tools meant to communicate complex relationships or ideas to a lay audience can easily be misinterpreted or misused. What I have found most helpful is when we do sophisticated analytical work with regression analysis and other things, and then Tableau is used to communicate relationships that we have already proven scientifically exist — as opposed to starting with a visualization and suggesting that a relationship exists or asserting that one does that might not."
Data Governance
Chambers said sharing data across areas of the university for strategic planning also leads to the need for a data governance effort. "We have a data governance committee, but it is in its infancy. We haven't started building the architecture to support policies in terms of how people access data."
"We are trying to fight the battle on two fronts: We are making progress on having data-informed decisions, and starting to work on data governance," Dalton added, "but we can't afford to wait until the perfect system of data definitions and data stewards is in place. We have to do both."
