Visualizing University Business Intelligence & Analytics Programs
Earlier this year, we collected information about the business intelligence and analytics programs that are currently being offered by universities in the United States. Not surprisingly, there was an incredible amount of interest in the result of that data collection effort, with more than 1000 list downloads. However, when I made an attempt at visualizing that data, I realized that the data was far too limited and that we’d need a richer data set.
Several months on, I’m happy to announce that our University BI and Analytics Programs, Version 2.1 is completed, and that the data does indeed present itself nicely in a Tableau Public visualization. Please feel free to interact with this visualization, and if you’re curious about the process that was followed in terms of collecting the data or putting together the visualization, continue reading below.
Collecting the Data
There were a number of difficulties we faced in putting this research report together. One major hurdle was the lack of complete information. Some websites were very good at putting the essential information in an easy to find place; other websites were not so cooperative. The same can be said of the responsiveness of some university departments when we reached out to a few for clarification. This process is still ongoing, and this information should be taken with a grain of salt until every university has responded.
Two of the most important factors we decided to focus on for this report were tuition and credit hours. While most universities made this information readily available, others were more difficult to find. In the best case scenario, we could switch between 2 or 3 pages to find the necessary information. In many cases, however, we found ourselves digging through 10 or more just to find a few entries.
With each unique university program, it is difficult to align them based on like factors. Sometimes the information is not available, and sometimes the program has a unique offering which changes some of the commonalities you might hope to see across universities. Even information as essential as credit hours and tuition varies in terms of availability of clear statements on the website, amount between individual students, and other factors. For example, Harvard and the Global MBA degree at NYU don’t operate on credit hour systems. For that reason, we were unable to include a numerical value in our graphic representation for those universities. Also, some universities, such as Virginia Commonwealth, have many factors that go into the tuition price for a student. Therefore, the numbers presented represented offer only a rough estimate of the best case scenario an average student might encounter.
Visualizing the Data
For data visualizing the data, Tableau Public was a natural choice. Not only is Tableau Public an extremely useful tool for freely sharing visual information, it has multiple features like out-of-the-box maps, variable-based behavior, and URL actions that are ideal for surfacing this data.
Use minimal space, as in something that fits into a blog entry
Allow users to navigate based on a variable of their choosing. Some users may want to filter out online vs. not online programs, while others may wish to focus on programs that are offered through a specific school, like the business school, etc.
Make the dashboard extremely simple to use.
Visualization Building Blocks
The basic building blocks were outlined as the following:
A – Slicer Parameter: As stated in the visualization goals section, one of the goals was to allow users to select which variable is most important to them. This was handled by creating a parameter that shows all of the relevant variables, then adding a view which lists relevant items based on the parameter selected in real-time. You will notice that changing the slicer (1-Slicer) impacts just about everything else on the dashboard. This is because the colors in the map and scatter plot are based on the slicer. More importantly, a “super-legend” view (for lack of a better name) is placed on the left of the scatter plot. This view serves multiple purposes by:
Providing a legend for the scatter plot that is easy to correlate by making the colors the same as in the scatter plot.
Providing additional data points. For example, the number of programs is included inside each circle.
Allowing the user to make selections from within the list of items pertaining to a slicer. For example, if the user has selected “Online Offered”, they can now selected “Yes” from this view.
B – States (Geography) View: Because the educational programs are geography-based, and given that most viewers of this data would be inclined to search for programs within a specific location, I added a map view which would act as a filter for the dashboard (i.e. clicking a state causes the dashboard data to be filtered down to just that state).
C – View Type Parameter, Scatter Plat & List View: This would be the primary analytical view of data, with the cost of the program on one axis and the length for completing the program on the other. As an optional view for someone who is perhaps just trying to get specific details on a program and then jump to the program website, I also created a basic list view with a URL action.
Putting it all Together
The dashboard offers more certain options that aren’t necessarily understood by end-users immediately. In order to encourage the end user to try playing with the options, I ordered the basic interactive features and included the feature number (1-Slicer, 2-State, etc…) and placed them in the dashboard on a left to right and top to bottom basis. This gets the user thinking in terms of going through the options as a series of steps, raising the likelihood that they will interact with some of the more important features, for example changing the slicer or the view type. Additional “touch-ups” on the visualization included setting the transparency to 70% and adding a slight outline (however, I didn’t add transparency to the scatter plot because the marks are pretty small and need to stand out).
Taken as a whole, this dashboard should encourage users to explore university BI and Analytics programs from multiple angles. Interesting trends will emerge during this process. For example, there’s a pretty consistent relationship between cost and length of the programs (the longer the program, the more it costs) which is pretty obvious. Less obvious are observations like the following:
Part-time programs are about $17K less expensive than full-time programs on average. Programs which are offered online tend to be about $9K less expensive, on average.
Programs that are offered online tend to require 37% fewer hours to complete.
The most “typical” department for these programs is the Computer Science, but not by a wide margin.
Master of Science is the most common type of degree, by a very wide margin. These programs average 35 hours and tend to cost an average of $31K.
There are certainly many other interesting data points to uncover. We hope you enjoy working with this visualization and find the data meaningful. For further details on the university programs, you can view our report here. Please let us know if you have any feedback!
Jared Decker has over twelve years of experience in the IT industry and eight years of consulting experience focused exclusively on data warehousing and business intelligence. He has been instrumental in the successful delivery of many projects, enabling organizations to achieve improved operational effectiveness through the timely availability of critical decision-making information. His breadth of experience entails everything from business data analysis through design and system implementation. He received a B.S. in Management Information Systems from the University of Tampa and an M.B.A. from the University of Houston. Mr. Decker holds several technical certifications related to database and corporate performance management systems.
You can subscribe to our RSS feed.