Guest Column: March 2020
DATA SCIENCE, PLAN B.
How today’s quick adaptation to COVID-19 mirrors the teaching of data science itself
Stephen Addison, Ph.D.
Dean, College of Natural Sciences and Mathematics University of Central Arkansas
AT UCA, WE are primarily a face-to-face campus, so transforming ourselves to an online university for the coronavirus is a big change for us. One thing I would share with the ACDS audience is that what we've had to do to get totally online strikes me as a microcosm of what we're having to do with data science—because with both we’re having to bring people at different levels up to speed. In the case of converting to online, we had to do it in days, rather than months or years, but it required many of the same sorts of meetings, the same sorts of discussions. Obviously, data science is newer to most people than learning online is, but in many ways it's the same skill set that was brought to bear.
Growing up in the United Kingdom, I came up in the older educational system where people were selected at an early age to go into university preparation courses. I passed those exams when I was 11, went to a school that had computers in the late 1960s, have used all sorts of computers, started with punch paper and cards, and essentially have worked with computers in all of their aspects throughout my career as a physicist, but also as a data analyst. I've also been in the administrative leadership of UCA’s College of Natural Science and Mathematics since it was founded in late 1994. As Dean for the last eight years, I've worked hard at making this a cutting-edge college. My guiding thought has always been, What should the college be doing to put ourselves in the best shape to be a university that succeeds and prospers into the future? There were a number of areas that I thought we should go into, including data science, cybersecurity, engineering, physics.
One of the first things I did as Dean was to challenge all of my departments to develop a data science effort, because I knew 10 years ago that data science was going to impact all disciplines. My mathematics and computer science departments stepped up and developed tracks within their degree programs, and today we have all of the pieces of a standalone degree program, particularly when we combine what's available throughout the College of Business. We also have strength in bioinformatics and biology, and even geography. Geography was in the College of Liberal Arts, but I recognized that geographic information systems were going to be central to geography in the future, and those information systems and spatial computing are a big part of data science. After many negotiations with the Dean of Liberal Arts, we were able to move the geography department into the College of Natural Sciences and Mathematics, where I believe the future of geography lies.
At this point, the only thing on my list that I haven't managed to implement is food safety— I think that will be an important area in the future as well. But in the computer science and data science and cybersecurity areas, we’ve been able to build the programs and add faculty, bringing in extraordinary expertise.
And now, suddenly, all of us have to do what we do online.
AT UCA, WE do regularly offer some online courses, but the number in science and mathematics is relatively small because you can't really do effective laboratories online. There are certain things that you can do, but until this pandemic, many of our faculty hadn't made the switch. Therein lay an adventure, and I’m proud to say that we accomplished our mission rather quickly. I’m writing this on Wednesday, March 18. We basically closed all our classes for two days, Friday the 13th and Monday the 16th, and were up and running as of Tuesday, March 17. Prior to this forced hiatus, we were offering a few hundred online classes. Now we’re offering more than 2700.
Among our faculty that weren't ready, there was a lot of worry. As a leader, I don't ask people to do things that I can't or won’t do myself, so I keep up to date in all aspects. And while I hadn't done much teaching online, I've used lots of hybrid elements in my classes—I know and understand the tools. So, along with my computer technician, I held a session for all of our college faculty: You’ve got to be teaching online tomorrow—what do you do?
We talked about the simplest tools and how we could use them. One such tool is Google Meet. All of our students have access to it and it's very easy to use, so we can teach that way. Then we talked about actually providing lesson content in an online manner. Rule #1: Don’t make an hour-long video, because people won't watch for that long. But aside from the question of content, people at remote sites may not have the bandwidth to download it. An hour of video may take 10 hours to download in some places, and that doesn't work.
Instead, I advised everybody to break up their classes into five- to seven-minute chunks, record them through something like QuickTime or VLC, and then post those videos to YouTube. In this uncertain world, one thing we know for sure is that all of our students watch YouTube. So, if you post to YouTube, they can find it, they can watch it, it streams wherever they are, and it also has automatically closed captions. Some people will have trouble understanding a video, just like in class we provide note takers for those students. But with a video on YouTube, they would be able to click on the closed caption button and get an automatic transcript of what was being said. We meet our Americans With Disabilities Act responsibilities if we post through YouTube.
One of the big mistakes people make with online teaching is thinking that they need to give the exact in-person teaching experience that students get in a real classroom. But no, students don’t have to see the professor—they just need to see the content and hear the professor explain it. As educators, we have to meet the same goals, but we do it in different ways. That assurance is very important to our students—they have to know that we’ll meet the goals of their courses, will provide them the tools to succeed, and that in addition to online classes they’ll have video conferencing opportunities, online office hours, whatever they need. If students are having an issue, I can connect with them in a Google Meet session, or share my desktop and talk them through the problem. Just because we’re teaching remotely doesn’t mean we’re “remote.”
I trust that this crisis will be over someday, and, when it is, we’ll be well positioned to continue using online teaching somewhat more than we had before. Online is a good add-on and provides additional opportunities. If you're teaching data science, you’re talking about using tools, and that works well online. I can fire up one of those tools, share my screen, and talk to students about the sorts of things I want them to do. And I can actually record the session so they can play it back afterwards. So, it varies between subjects, but for the hard laboratory sciences, which are like apprenticeships, there's no real substitute for the hands-on experience.
I'm teaching a course on electronics currently, and normally what I would do is have students build the circuits and learn by trial and error how to get a circuit working. I can show them, online, how to build a circuit using a simulator. But the problem with using a simulator is that most of them tell you when you've done something wrong, so if you have an open circuit you don’t get the valuable experience of puzzling over it for a couple of hours thinking this should be working, and figuring it out for yourself. So we’ll know that this particular crop of students did things virtually rather than actually hands-on, so they'll need to come back and brush up with some face-to-face help.
I BEGAN THIS piece by drawing a parallel between UCA’s process of converting to online learning and the process of teaching the data sciences. For many months here in Arkansas, a number of people and groups have been working on the broader goal of making a data science curriculum available to students all over the state. Included in this ongoing discussion are Governor Hutchinson and the members of his Computer Science and Cybersecurity Task Force; leaders from the Department of Higher Education; the people at ACDS; and educators from two- and four-year schools from the four corners of Arkansas. Ultimately, our goal is to bring everyone up to speed on why data science is key to the economic future of our state, and then to agree on how best to share the duties of preparing our young people for careers in the data sciences.
With the help and encouragement of Bill Yoder at ACDS, Dr. Karl Schubert at the University of Arkansas in Fayetteville and I have taken the lead in explaining the “whats” and “whys” of a statewide data science program via a series of Data Science Workshops, and we were to speak to the “hows” in another workshop previously scheduled for April, now postponed. While the present pandemic will delay us, we will eventually make this vision a reality. But as with educating professors to stage their courses online, this goal of a statewide data science program requires some teaching.
It is important that we not reinvent the wheel at every school across Arkansas. What we foresee is a system in which students from everywhere in the state can get a core data science education and expertise, but campuses would necessarily specialize in different areas that are appropriate to the regions they’re in. At the University of Arkansas at Fayetteville, for instance, there’s a big effort in power plants and power systems, with the accompanying emphasis on cybersecurity. And logistics is important in Northwest Arkansas, with all their distribution centers. Here in Central Arkansas, we have the Conway headquarters of Acxiom, with its specialty in business services, mailing lists, and in Little Rock, First Orion is working on security and other applications for cell phones. So the particular needs differ according to what area industries are focusing on. The idea is to provide different programs to build on the core, but to specialize in areas that are useful to industries in a particular area.
Another area of discussion and, hopefully, persuasion is the matter of what tools we should be using statewide. Some people are still favoring tools like SPSS and SAS, tools that were big in their day but are not the tools that modern data science is being built with. Today, the prevalent tools are a language called R and a language called Python, and it goes back and forth as to which is the more popular language. But those two are overwhelmingly the languages that data science companies need us to be exposed to. Our goal is to make sure everybody knows what the basic tools are, and to provide opportunities for them to learn how to use them.
Then we have to agree on the core courses that everybody needs to take. We can offer them in many places, but we also recognize that a small two-year school may not be able to offer all the intro classes. So now we need to provide a way to share those classes through what I term a “distributed department.” Let’s say I've got a class of 18 students at UCA. We can bring in three additional students from a two-year school online without any additional overhead for the instructor. These are the sorts of things Karl Schubert and I are going around talking to people about.
There are still some people saying, “Oh, no, we can't do that—we’ve got to think about the financial aspects.” And what I say is, we can't afford not to do it, because we have an opportunity here. We're maybe a little behind some of the states on the east and west coasts, but we’re ahead of most other states in the country. So if we do this now, we can be a leader. What I like to say is, Arkansas can be the center of data science for flyover country. We can be that hook. But if we say, “We can't do this, nobody else has done this before,” you can bet that somebody else will do it. If we do it and continue to lead, we'll all prosper. Because we won't just have students from Arkansas, we'll have students, in person or online, from all those other states who haven't been doing these things.
So whether we’re dealing with our physical health against coronavirus, or our economic health against narrow-mindedness, we need to be thinking outside of the box. Our mode needs to be, Let's see how we can cooperate, not how we can compete. Karl Schubert at UA Fayetteville and I at UCA are examples of this. We both have programs, and we both want students in those programs. But we know that by sharing things and cooperating, we'll both be better off. We need to get everybody else in that same headspace. Arkansas is only going to lead if we're all on the same page, all working together.